diff --git a/.github/workflows/model-metadata-upkeep.yml b/.github/workflows/model-metadata-upkeep.yml new file mode 100644 index 0000000000..07016276d0 --- /dev/null +++ b/.github/workflows/model-metadata-upkeep.yml @@ -0,0 +1,163 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +name: Model metadata upkeep + +on: + schedule: + # Weekly, offset from the hour to reduce peak-time scheduling delays. The + # snapshot is a build input a human reviews, so a nightly cadence would + # only stack five near-identical pull requests against one week of upstream + # movement, on runners the whole foundation shares. + - cron: '41 6 * * 1' + workflow_dispatch: + +permissions: + contents: read + +concurrency: + group: model-metadata-upkeep + cancel-in-progress: false + +jobs: + refresh: + # A fork inherits the schedule but owns neither the branch this pushes nor + # the pull request it opens. + if: github.repository == 'apache/maka' + runs-on: ubuntu-latest + timeout-minutes: 30 + permissions: + contents: write + pull-requests: write + steps: + - name: Check out the repository + uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 + with: + persist-credentials: false + + - name: Set up Node.js + uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0 + with: + node-version: '24' + cache: npm + + - name: Install dependencies + run: npm ci --ignore-scripts + + - name: Report snapshot drift against models.dev + # Exit 2 is "upstream moved", which is the expected outcome and the + # reason this step exists. Any other non-zero status is the command + # itself failing, and the job stops on it. + run: | + npm run --silent check:model-metadata-drift > "$RUNNER_TEMP/drift.txt" || { + status=$? + cat "$RUNNER_TEMP/drift.txt" + [ "$status" -eq 2 ] || exit "$status" + } + + - name: Publish the drift report + run: | + { + echo '### models.dev drift' + echo '' + echo '```text' + cat "$RUNNER_TEMP/drift.txt" + echo '```' + } >> "$GITHUB_STEP_SUMMARY" + + # --accept-upstream-removals, because the review seat this job is built + # around is the draft pull request below. A person still inspects every + # removal and still decides, in the diff, with the drift report in the + # body; refusing here would only make the job red every week, since it + # cannot rerun itself the way the acknowledgement asks a human to. + - name: Refresh the snapshot from models.dev + run: npm run refresh:model-metadata -- --accept-upstream-removals + + - name: Verify the regenerated outputs + run: npm run check:model-metadata + + - name: Detect a snapshot change + id: change + run: | + if git diff --quiet -- scripts/model-metadata/models-dev-api.snapshot.json; then + echo 'changed=false' >> "$GITHUB_OUTPUT" + else + echo 'changed=true' >> "$GITHUB_OUTPUT" + fi + + - name: Open the review pull request + if: steps.change.outputs.changed == 'true' + env: + BRANCH: automation/model-metadata-refresh + GH_TOKEN: ${{ github.token }} + TITLE: 'chore(model-metadata): refresh the models.dev snapshot' + run: | + { + echo '## Summary' + echo '' + echo 'Scheduled `refresh:model-metadata` run. The snapshot is the build' + echo 'input for the bundled model catalog; this only moves it to what' + echo 'models.dev serves today.' + echo '' + echo 'Refs #4398' + echo '' + echo '## Verification' + echo '' + echo 'The workflow ran `refresh:model-metadata` and `check:model-metadata`' + echo 'before opening this. Drift against upstream at refresh time:' + echo '' + echo '```text' + cat "$RUNNER_TEMP/drift.txt" + echo '```' + } > "$RUNNER_TEMP/pr-body.md" + git config user.name 'Apache Maka' + git config user.email 'commits@maka.apache.org' + git switch -c "$BRANCH" + git add scripts/model-metadata/models-dev-api.snapshot.json + git commit -m "$TITLE" + # The token travels in a header rather than the remote URL, which git + # echoes back in its own error messages. + AUTH="$(printf 'x-access-token:%s' "$GH_TOKEN" | base64 | tr -d '\n')" + REMOTE="https://github.com/${GITHUB_REPOSITORY}" + TIP="$(git -c "http.extraheader=Authorization: Basic $AUTH" \ + ls-remote "$REMOTE" "refs/heads/$BRANCH" | cut -f1)" + if [ -n "$TIP" ]; then + # This branch only ever carries commits this job wrote. Anything + # else is a person working on the open pull request, and a force + # push would erase it. + git -c "http.extraheader=Authorization: Basic $AUTH" \ + fetch --depth=1 "$REMOTE" "refs/heads/$BRANCH" + if [ "$(git log -1 --format=%s FETCH_HEAD)" != "$TITLE" ]; then + echo "::error::$BRANCH carries a commit this workflow did not write; refusing to overwrite it." + exit 1 + fi + git -c "http.extraheader=Authorization: Basic $AUTH" push \ + "--force-with-lease=refs/heads/$BRANCH:$TIP" "$REMOTE" "HEAD:refs/heads/$BRANCH" + else + git -c "http.extraheader=Authorization: Basic $AUTH" push \ + "$REMOTE" "HEAD:refs/heads/$BRANCH" + fi + # A plain existence lookup also succeeds for a closed pull request, + # which would leave a maintainer's decision to close one silently + # disabling this job forever. Only an open one is one to update. + if [ "$(gh pr list --head "$BRANCH" --state open --json number --jq 'length')" -gt 0 ]; then + gh pr edit "$BRANCH" --body-file "$RUNNER_TEMP/pr-body.md" + echo "Updated the open pull request on $BRANCH." + else + gh pr create --draft --base main --head "$BRANCH" \ + --title "$TITLE" --body-file "$RUNNER_TEMP/pr-body.md" + fi diff --git a/package.json b/package.json index 08a3131b68..0cd0891641 100644 --- a/package.json +++ b/package.json @@ -62,7 +62,7 @@ "release:asf:source": "node scripts/asf-source-release.mjs create", "release:asf:verify": "node scripts/asf-source-release.mjs verify", "release:asf:sign": "node scripts/asf-source-release.mjs sign", - "check:asf-source": "npm run check:model-metadata && node --test scripts/asf-source-release.test.mjs scripts/asf-source-workflow-policy.test.mjs scripts/asf-license-headers.test.mjs scripts/source-legal-inventory.test.mjs scripts/sync-model-metadata.test.mjs", + "check:asf-source": "npm run check:model-metadata && node --test scripts/asf-source-release.test.mjs scripts/asf-source-workflow-policy.test.mjs scripts/asf-license-headers.test.mjs scripts/model-metadata-upkeep-workflow-policy.test.mjs scripts/source-legal-inventory.test.mjs scripts/sync-model-metadata.test.mjs", "check:asf-npm": "node --test scripts/asf-npm-workflow-policy.test.mjs", "check:product-release-identity": "node scripts/product-release-identity.mjs", "package:cli:macos-arm64": "node scripts/package-macos-arm64-cli.mjs", @@ -92,6 +92,7 @@ "sync:model-metadata": "node scripts/sync-model-metadata.mjs", "refresh:model-metadata": "node scripts/sync-model-metadata.mjs --refresh", "check:model-metadata": "node scripts/sync-model-metadata.mjs --check", + "check:model-metadata-drift": "node scripts/sync-model-metadata.mjs --drift", "generate:bundled-skills": "node scripts/gen-bundled-skill-catalog.mjs", "computer-use": "node scripts/computer-use.mjs", "windows:inventory": "node --test scripts/windows-test-inventory.test.mjs && node scripts/windows-test-inventory.mjs --check", diff --git a/packages/core/src/__tests__/model-catalog.test.ts b/packages/core/src/__tests__/model-catalog.test.ts index afb0fff804..b6a6d7574a 100644 --- a/packages/core/src/__tests__/model-catalog.test.ts +++ b/packages/core/src/__tests__/model-catalog.test.ts @@ -112,19 +112,28 @@ test('a declared output modality without text rules a model out of chat', () => modelSource: 'fetched' as const, }; assert.deepEqual(verdict(audioOnly), { ok: false }); -}); -test('an empty output modality list is not evidence against chat', () => { - // `modalities.output` is typed to text, image, and audio, so a video model's - // real output has no representation and serializes as `[]` — the same shape - // a generator bug would produce. Blocking on it would be guessing. - const video = { + // Video-only says exactly what the other two say. It could not be read at + // all until `modalities.output` could carry the value. + const videoOnly = { providerType: 'google' as const, defaultModel: 'gemini-omni-flash-preview', models: [{ id: 'gemini-omni-flash-preview' }], modelSource: 'fetched' as const, }; - assert.deepEqual(verdict(video), { ok: true }); + assert.deepEqual(verdict(videoOnly), { ok: false }); +}); + +test('an empty output modality list is not evidence against chat', () => { + // A provider that declared no output modality and a generator bug that + // dropped them produce the same shape. Blocking on it would be guessing. + const undeclared = { + providerType: 'openai-compatible' as const, + defaultModel: 'relay-quiet', + models: [{ id: 'relay-quiet', modalities: { input: ['text' as const], output: [] } }], + modelSource: 'fetched' as const, + }; + assert.deepEqual(verdict(undeclared), { ok: true }); }); test('an explicit chat capability outranks the declared output modality', () => { diff --git a/packages/core/src/__tests__/model-metadata.test.ts b/packages/core/src/__tests__/model-metadata.test.ts index 8f9e01cfe0..2696510d85 100644 --- a/packages/core/src/__tests__/model-metadata.test.ts +++ b/packages/core/src/__tests__/model-metadata.test.ts @@ -138,7 +138,6 @@ describe('deepseek v4 flash vision exp metadata regression', () => { metadata.description, 'Experimental DeepSeek V4 Flash model for image understanding and multimodal agent tasks', ); - assert.equal(metadata.docsUrl, 'https://api-docs.deepseek.com/guides/vision/'); assert.equal(metadata.contextWindow, 1_000_000); assert.equal(metadata.maxOutputTokens, 384_000); assert.equal(metadata.structuredOutput, true); diff --git a/packages/core/src/__tests__/runtime-policy-codec.test.ts b/packages/core/src/__tests__/runtime-policy-codec.test.ts index 06bdd1328c..d6aba85a60 100644 --- a/packages/core/src/__tests__/runtime-policy-codec.test.ts +++ b/packages/core/src/__tests__/runtime-policy-codec.test.ts @@ -511,6 +511,19 @@ test('normalizes extended model facts used by the runtime host catalog', () => { }); }); +test('carries the video and pdf modalities models.dev declares', () => { + const modalities = { + input: ['text', 'image', 'video'], + output: ['text', 'pdf', 'video'], + }; + const result = normalizeConnectionModelDiscoveryResult({ + models: [{ id: 'custom-model', modalities }], + source: 'fetched', + fetchedAt: 42, + }); + assert.deepEqual(result.models[0], { id: 'custom-model', modalities }); +}); + test('rejects sparse model modality arrays', () => { assert.throws( () => diff --git a/packages/core/src/llm-connections.ts b/packages/core/src/llm-connections.ts index 5a4a839058..1d6c605d19 100644 --- a/packages/core/src/llm-connections.ts +++ b/packages/core/src/llm-connections.ts @@ -86,6 +86,20 @@ export type ConnectionAuth = | { kind: 'oauth_token'; oauthToken: string; expiresAt?: number } | { kind: 'none' }; +/** + * The modalities a model may declare on either side. Every validator that + * admits a modality reads this one set: a decoder, an overlay normalizer, and + * a live-fetch reader each holding their own copy is how one of them stayed a + * catalog behind the others. + */ +export type ModelModality = 'text' | 'image' | 'audio' | 'pdf' | 'video'; + +const MODEL_MODALITIES: readonly ModelModality[] = ['text', 'image', 'audio', 'pdf', 'video']; + +export function isModelModality(value: unknown): value is ModelModality { + return MODEL_MODALITIES.includes(value as ModelModality); +} + export interface ModelInfo { id: string; displayName?: string; @@ -116,8 +130,8 @@ export interface ModelInfo { }; /** Multimodal input/output support from provider catalog metadata. */ modalities?: { - input: Array<'text' | 'image' | 'audio' | 'pdf'>; - output: Array<'text' | 'image' | 'audio'>; + input: ModelModality[]; + output: ModelModality[]; }; /** * Read-time provenance for values overlaid from model-facts.json. This is diff --git a/packages/core/src/model-catalog.ts b/packages/core/src/model-catalog.ts index 12d2609ac4..45e5db3c0d 100644 --- a/packages/core/src/model-catalog.ts +++ b/packages/core/src/model-catalog.ts @@ -515,10 +515,9 @@ function displayNameForKnownModel( * never set `capabilities.imageGeneration` for any of them, so the capability * check below could not fire on bundled data. * - * An EMPTY list is not evidence. `modalities.output` is typed to text, image, - * and audio, so a video model's real output has no representation and - * serializes as `[]` — the same shape a future generator bug would produce. - * Only a non-empty list says something, and what it says is what it lists. + * An EMPTY list is not evidence. A provider that declared no output modality + * and a generator bug that dropped them produce the same shape. Only a + * non-empty list says something, and what it says is what it lists. */ function declaresNoTextOutput(model: ModelInfo): boolean { const output = model.modalities?.output; diff --git a/packages/core/src/model-facts.ts b/packages/core/src/model-facts.ts index 99f9ecbbef..7b166481e9 100644 --- a/packages/core/src/model-facts.ts +++ b/packages/core/src/model-facts.ts @@ -18,6 +18,7 @@ */ import { providerDefaultsOf, type ProviderType } from './provider-registry.js'; +import { isModelModality } from './llm-connections.js'; import type { ModelFactField, ModelInfo } from './llm-connections.js'; import { CONNECTION_CATALOG_MAX_MODELS_PER_CONNECTION, @@ -193,8 +194,8 @@ function normalizeModalities(value: unknown): NonNullable( return [...new Set(entries)]; } -function isModality(value: unknown): value is 'text' | 'image' | 'audio' | 'pdf' { - return value === 'text' || value === 'image' || value === 'audio' || value === 'pdf'; -} -function isOutputModality(value: unknown): value is 'text' | 'image' | 'audio' { - return value === 'text' || value === 'image' || value === 'audio'; -} function isPositiveBoundedInteger(value: unknown): value is number { return ( typeof value === 'number' && diff --git a/packages/core/src/model-metadata.ts b/packages/core/src/model-metadata.ts index 4c8268be63..81366ad80f 100644 --- a/packages/core/src/model-metadata.ts +++ b/packages/core/src/model-metadata.ts @@ -28,7 +28,6 @@ export interface ModelMetadata { displayName?: string; description?: string; lifecycle?: 'active' | 'beta' | 'alpha' | 'deprecated' | 'retired'; - docsUrl?: string; contextWindow?: number; inputLimit?: number; maxOutputTokens?: number; @@ -242,8 +241,6 @@ const SILICONFLOW_MODEL_OVERRIDES: Record = Object.fromEn .map(([id]) => [id, { capabilities: { chat: true } }]), ); -const VOLCENGINE_CODING_PLAN_DOCS = 'https://www.volcengine.com/docs/82379/1925114'; -const VOLCENGINE_AGENT_PLAN_DOCS = 'https://www.volcengine.com/docs/82379/2366394'; const VOLCENGINE_CODING_PLAN_MODEL_METADATA: Record = { 'ark-code-latest': planModel('Ark Code Latest', false), 'doubao-seed-2.0-code': planModel('Doubao Seed 2.0 Code', true), @@ -351,7 +348,6 @@ const STATIC_MODEL_METADATA: Partial decodeModelInputModality(entry)); - const output = Array.from(item.output, (entry) => decodeModelOutputModality(entry)); + const input = Array.from(item.input, (entry) => decodeModelModality(entry, 'input')); + const output = Array.from(item.output, (entry) => decodeModelModality(entry, 'output')); return { input, output }; } -function decodeModelInputModality(value: unknown): 'text' | 'image' | 'audio' | 'pdf' { - const modality = stringValue(value, 'connection model input modality', 16); - if (modality !== 'text' && modality !== 'image' && modality !== 'audio' && modality !== 'pdf') { - throw domainError('connection model input modality is invalid'); - } - return modality; -} - -function decodeModelOutputModality(value: unknown): 'text' | 'image' | 'audio' { - const modality = stringValue(value, 'connection model output modality', 16); - if (modality !== 'text' && modality !== 'image' && modality !== 'audio') { - throw domainError('connection model output modality is invalid'); +// Both directions accept the same set. models.dev declares video on either +// side and pdf on both, so splitting them again only invites one direction to +// drift behind the catalog it decodes. +function decodeModelModality(value: unknown, direction: 'input' | 'output'): ModelModality { + const modality = stringValue(value, `connection model ${direction} modality`, 16); + if (!isModelModality(modality)) { + throw domainError(`connection model ${direction} modality is invalid`); } return modality; } diff --git a/packages/runtime-host/src/protocol/index.ts b/packages/runtime-host/src/protocol/index.ts index 8aff63a16a..0a7784ff03 100644 --- a/packages/runtime-host/src/protocol/index.ts +++ b/packages/runtime-host/src/protocol/index.ts @@ -95,7 +95,13 @@ export const RUNTIME_HOST_REGISTRATION_SCHEMA_VERSION = 1 as const; export const RUNTIME_HOST_PROTOCOL_VERSION = 0 as const; // Increment when the same protocol version no longer guarantees safe Client-Host // interoperability. Mismatches are rejected before domain commands are admitted. -export const RUNTIME_HOST_COMPATIBILITY_EPOCH = 87 as const; +export const RUNTIME_HOST_COMPATIBILITY_EPOCH = 88 as const; +// 88: Catalog model modalities admit video on either side and pdf as output. +// models.dev declares both, and the modality decoder rejects any value it does +// not name, so a newer Host describing such a model fails an older client's +// catalog decode outright rather than losing one field. The handshake keeps +// that pairing from forming; a newer client simply never sees the new values +// from an older Host. // 87: The connection catalog projects each model as the Host resolved it — // a `catalog_entry` item per model, counted by the connection header. Clients // render those entries instead of merging the stored row against their own diff --git a/packages/runtime/src/__tests__/computer-use-provider-protocol.test.ts b/packages/runtime/src/__tests__/computer-use-provider-protocol.test.ts index 70ae04705d..8923a1e74b 100644 --- a/packages/runtime/src/__tests__/computer-use-provider-protocol.test.ts +++ b/packages/runtime/src/__tests__/computer-use-provider-protocol.test.ts @@ -199,7 +199,7 @@ describe('Anthropic-compatible Computer Use product loops', () => { auth: 'x-api-key', expectedAuth: 'test-key', expectedThinking: undefined, - expectedWireOutputLimit: 128_000, + expectedWireOutputLimit: 512_000, apiProtocol: undefined, }, { diff --git a/packages/runtime/src/__tests__/model-factory-thinking.test.ts b/packages/runtime/src/__tests__/model-factory-thinking.test.ts index 6a33ad80f2..4762c2d72c 100644 --- a/packages/runtime/src/__tests__/model-factory-thinking.test.ts +++ b/packages/runtime/src/__tests__/model-factory-thinking.test.ts @@ -684,14 +684,15 @@ describe('buildProviderOptions: thinking level', () => { }); test('Tencent Token Plan sends its documented reasoning effort under the stable provider namespace', () => { - assert.deepEqual( - [...thinkingVariantsForModel('tencent-token-plan', 'hy3')], - ['low', 'medium', 'high'], - ); + assert.deepEqual([...thinkingVariantsForModel('tencent-token-plan', 'hy3')], ['off', 'high']); assert.deepEqual(buildProviderOptions(conn('tencent-token-plan'), 'hy3', 'high'), { tencentTokenPlan: { reasoningEffort: 'high' }, }); - assert.deepEqual(buildProviderOptions(conn('tencent-token-plan'), 'hy3', 'off'), {}); + // The plan documents an explicit no-reasoning value, so off names it rather + // than dropping the knob and leaving the provider to pick. + assert.deepEqual(buildProviderOptions(conn('tencent-token-plan'), 'hy3', 'off'), { + tencentTokenPlan: { reasoningEffort: 'none' }, + }); }); test('Vercel Gateway sends reasoning effort under its stable namespace and exact model id', () => { diff --git a/packages/runtime/src/model-fetcher.ts b/packages/runtime/src/model-fetcher.ts index 14de28638c..e886114c3e 100644 --- a/packages/runtime/src/model-fetcher.ts +++ b/packages/runtime/src/model-fetcher.ts @@ -21,9 +21,11 @@ import { PROVIDER_REGISTRY, providerFallbackModelIds, effectiveBaseUrl, + isModelModality, providerAuthSupportsApiKey, type LlmConnection, type ModelInfo, + type ModelModality, } from '@maka/core/llm-connections'; import { generalizedErrorMessage } from '@maka/core/redaction'; import { @@ -883,12 +885,9 @@ function providerObjectArray( * non-string — is dropped rather than guessed at, so an unrecognized list * reads as "said nothing" instead of "said not text". */ -function knownOutputModalities(declared: readonly unknown[] | undefined): string[] { +function knownOutputModalities(declared: readonly unknown[] | undefined): ModelModality[] { if (declared === undefined) return []; - return declared.filter( - (value): value is 'text' | 'image' | 'audio' => - value === 'text' || value === 'image' || value === 'audio', - ); + return declared.filter(isModelModality); } function assertOptionalArray( diff --git a/scripts/ci-test-plan.mjs b/scripts/ci-test-plan.mjs index 149dd30152..b66320136c 100644 --- a/scripts/ci-test-plan.mjs +++ b/scripts/ci-test-plan.mjs @@ -118,6 +118,7 @@ const CLI_PACKAGE_FILES = new Set([ const ASF_SOURCE_FILES = new Set([ '.github/workflows/asf-source-candidate.yml', + '.github/workflows/model-metadata-upkeep.yml', 'DISCLAIMER-WIP', 'LICENSE', 'NOTICE', @@ -131,6 +132,7 @@ const ASF_SOURCE_FILES = new Set([ 'scripts/asf-source-release.mjs', 'scripts/asf-source-release.test.mjs', 'scripts/asf-source-workflow-policy.test.mjs', + 'scripts/model-metadata-upkeep-workflow-policy.test.mjs', 'scripts/model-metadata/models-dev-api.snapshot.json', 'scripts/source-legal-inventory.test.mjs', 'scripts/sync-model-metadata.mjs', diff --git a/scripts/model-metadata-upkeep-workflow-policy.test.mjs b/scripts/model-metadata-upkeep-workflow-policy.test.mjs new file mode 100644 index 0000000000..8e82644b2a --- /dev/null +++ b/scripts/model-metadata-upkeep-workflow-policy.test.mjs @@ -0,0 +1,116 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +import assert from 'node:assert/strict'; +import { readFile } from 'node:fs/promises'; +import { test } from 'node:test'; +import { parse } from 'yaml'; + +async function readUpkeepWorkflow() { + return parse( + await readFile( + new URL('../.github/workflows/model-metadata-upkeep.yml', import.meta.url), + 'utf8', + ), + ); +} + +async function readRootScripts() { + return JSON.parse(await readFile(new URL('../package.json', import.meta.url), 'utf8')).scripts; +} + +function stepNamed(workflow, name) { + const step = workflow.jobs.refresh.steps.find((candidate) => candidate.name === name); + assert.ok(step, `missing step: ${name}`); + return step; +} + +test('upkeep runs on a weekly schedule and never against a pull request', async () => { + const workflow = await readUpkeepWorkflow(); + assert.deepEqual(Object.keys(workflow.on).toSorted(), ['schedule', 'workflow_dispatch']); + assert.equal(workflow.on.schedule.length, 1); + const [minute, hour, dayOfMonth, month, dayOfWeek] = workflow.on.schedule[0].cron.split(' '); + // A single weekday with a fixed hour is the weekly cadence. Runners are a + // shared foundation resource and every daily run costs a queue slot. + assert.match(dayOfWeek, /^[0-6]$/u); + assert.deepEqual([dayOfMonth, month], ['*', '*']); + assert.doesNotMatch(minute, /^[*0]$/u); + assert.match(hour, /^\d+$/u); +}); + +test('only the scheduled job writes, and only on the canonical repository', async () => { + const workflow = await readUpkeepWorkflow(); + assert.deepEqual(workflow.permissions, { contents: 'read' }); + assert.deepEqual(workflow.jobs.refresh.permissions, { + contents: 'write', + 'pull-requests': 'write', + }); + assert.match(workflow.jobs.refresh.if, /github\.repository == 'apache\/maka'/u); + assert.equal(Object.keys(workflow.jobs).length, 1); + assert.equal(stepNamed(workflow, 'Check out the repository').with['persist-credentials'], false); +}); + +test('the refresh reports drift before it writes, and tolerates only drift', async () => { + const workflow = await readUpkeepWorkflow(); + const order = [ + 'Report snapshot drift against models.dev', + 'Refresh the snapshot from models.dev', + ].map((name) => workflow.jobs.refresh.steps.findIndex((step) => step.name === name)); + assert.ok(order.every((position) => position >= 0)); + assert.ok(order[0] < order[1]); + const drift = stepNamed(workflow, 'Report snapshot drift against models.dev'); + // A blanket continue-on-error would hide the command failing outright. Only + // the documented drift status is tolerated. + assert.equal(drift['continue-on-error'], undefined); + assert.match(drift.run, /\[ "\$status" -eq 2 \] \|\| exit "\$status"/u); +}); + +test('the pull request is opened for review and never merged by the job', async () => { + const workflow = await readUpkeepWorkflow(); + const open = stepNamed(workflow, 'Open the review pull request'); + assert.match(open.run, /gh pr create --draft/u); + assert.doesNotMatch(JSON.stringify(workflow), /gh pr merge|--auto\b|--admin\b/u); + // A closed pull request must not read as one to update; gh pr view says it + // does, so only an open-state listing decides. + assert.doesNotMatch(open.run, /gh pr view/u); + assert.match(open.run, /gh pr list --head "\$BRANCH" --state open/u); +}); + +test('the branch this job pushes never loses a commit someone else wrote', async () => { + const workflow = await readUpkeepWorkflow(); + const open = stepNamed(workflow, 'Open the review pull request'); + assert.doesNotMatch(open.run, /push\s+--force\b|push\s+-f\b/u); + assert.match(open.run, /--force-with-lease=refs\/heads\/\$BRANCH:\$TIP/u); + // The token belongs in a header. A remote URL carrying it is echoed back by + // git's own error messages. + assert.doesNotMatch(JSON.stringify(workflow), /x-access-token:\$\{?GH_TOKEN/u); +}); + +test('the drift check stays out of the checks that run on every pull request', async () => { + const scripts = await readRootScripts(); + assert.equal( + scripts['check:model-metadata-drift'], + 'node scripts/sync-model-metadata.mjs --drift', + ); + // Reaching models.dev is not a precondition for reviewing a pull request. + for (const gate of ['check:asf-source', 'check:release']) { + assert.doesNotMatch(scripts[gate], /check:model-metadata-drift|--drift/u); + } + assert.match(scripts['check:asf-source'], /model-metadata-upkeep-workflow-policy\.test\.mjs/u); +}); diff --git a/scripts/model-metadata/models-dev-api.snapshot.json b/scripts/model-metadata/models-dev-api.snapshot.json index 34667f1fcc..c7c5161db3 100644 --- a/scripts/model-metadata/models-dev-api.snapshot.json +++ b/scripts/model-metadata/models-dev-api.snapshot.json @@ -2,10 +2,12 @@ "formatVersion": 1, "sourceUrl": "https://models.dev/api.json", "origin": { - "kind": "generated-output-migration", - "commit": "729839ed8ada3e5498b0ad27e64fb42dfd283ae7" + "kind": "models-dev-response", + "retrievedAt": "2026-09-01T10:29:38.000Z", + "etag": "W/\"7559b5b3701539769901f1e20e89315d\"", + "responseSha256": "7559b5b3701539769901f1e20e89315d94975a7b289646590d6f854a21e12f17" }, - "projectionSha256": "6957e4d11c649fa77dc33474bd67390a28d290c994f34dabac23f539b5f59149", + "projectionSha256": "85b25f334b65929079ea6dab9b0a2a66f433e3d66710fcd1aadff5f611b37345", "projection": { "metadata": { "anthropic": { @@ -13,7 +15,6 @@ "displayName": "Claude Fable 5", "description": "Claude model for creative writing, analysis, and controlled agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -35,7 +36,6 @@ "displayName": "Claude Haiku 4.5 (latest)", "description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-02-28", @@ -55,7 +55,6 @@ "displayName": "Claude Haiku 4.5", "description": "Fast Claude model for responsive assistance, classification, and lightweight agents", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-02-28", @@ -75,7 +74,6 @@ "displayName": "Claude Opus 4.5 (latest)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-05", @@ -98,7 +96,6 @@ "displayName": "Claude Opus 4.5", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-05", @@ -121,7 +118,6 @@ "displayName": "Claude Opus 4.6", "description": "High-end Claude for difficult coding, planning, and slower expert reasoning", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05-31", @@ -144,7 +140,6 @@ "displayName": "Claude Opus 4.7", "description": "Stronger Opus tier for advanced software work and high-stakes reasoning", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -167,7 +162,6 @@ "displayName": "Claude Opus 4.8", "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01", @@ -190,7 +184,6 @@ "displayName": "Claude Opus 5", "description": "Strongest Claude Opus model for coding, agents, and professional work", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-05", @@ -213,7 +206,6 @@ "displayName": "Claude Sonnet 4.5 (latest)", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-07-31", @@ -233,7 +225,6 @@ "displayName": "Claude Sonnet 4.5", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-07-31", @@ -253,7 +244,6 @@ "displayName": "Claude Sonnet 4.6", "description": "Claude workhorse for coding agents, careful analysis, and production cost control", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -276,7 +266,6 @@ "displayName": "Claude Sonnet 5", "description": "Everyday Claude agent model for coding, planning, browsing, and general work", "lifecycle": "active", - "docsUrl": "https://docs.anthropic.com/en/docs/about-claude/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -302,7 +291,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -326,7 +314,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -348,7 +335,6 @@ "displayName": "QVQ Max", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -367,7 +353,6 @@ "displayName": "Qwen Flash", "description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -389,7 +374,6 @@ "displayName": "Qwen Max", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -408,7 +392,6 @@ "displayName": "Qwen-MT Plus", "description": "Translation model for multilingual conversion, localization, and cross-language workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 16384, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -427,7 +410,6 @@ "displayName": "Qwen-MT Turbo", "description": "Translation model for multilingual conversion, localization, and cross-language workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 16384, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -446,7 +428,6 @@ "displayName": "Qwen-Omni Turbo", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-04", @@ -457,7 +438,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -465,7 +446,6 @@ "displayName": "Qwen-Omni Turbo Realtime", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-04", @@ -484,7 +464,6 @@ "displayName": "Qwen Plus", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -506,7 +485,6 @@ "displayName": "Qwen Plus Character (Japanese)", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 8192, "maxOutputTokens": 512, "knowledgeCutoff": "2024-04", @@ -525,7 +503,6 @@ "displayName": "Qwen Turbo", "description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-04", @@ -547,7 +524,6 @@ "displayName": "Qwen-VL Max", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -566,7 +542,6 @@ "displayName": "Qwen-VL OCR", "description": "OCR model for extracting structured text from documents and screenshots", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 34096, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-04", @@ -585,7 +560,6 @@ "displayName": "Qwen-VL Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -604,7 +578,6 @@ "displayName": "Qwen2.5 14B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -623,7 +596,6 @@ "displayName": "Qwen2.5 32B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -642,7 +614,6 @@ "displayName": "Qwen2.5 72B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -661,7 +632,6 @@ "displayName": "Qwen2.5 7B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -680,7 +650,6 @@ "displayName": "Qwen2.5-Omni 7B", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-04", @@ -691,7 +660,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -699,7 +668,6 @@ "displayName": "Qwen2.5-VL 72B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -718,7 +686,6 @@ "displayName": "Qwen2.5-VL 7B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -737,7 +704,6 @@ "displayName": "Qwen3 14B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-04", @@ -759,7 +725,6 @@ "displayName": "Qwen3 235B-A22B", "description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -781,7 +746,6 @@ "displayName": "Qwen3 32B", "description": "Dense open Qwen model for self-hosted chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -803,7 +767,6 @@ "displayName": "Qwen3 8B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-04", @@ -825,7 +788,6 @@ "displayName": "Qwen3-ASR Flash", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 53248, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-04", @@ -844,7 +806,6 @@ "displayName": "Qwen3-Coder 30B-A3B Instruct", "description": "Smaller Qwen coder for efficient local agents and repo-level fixes", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -863,7 +824,6 @@ "displayName": "Qwen3-Coder 480B-A35B Instruct", "description": "Open Qwen coding heavyweight for repository reasoning and agentic engineering", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -882,7 +842,6 @@ "displayName": "Qwen3 Coder Flash", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -901,7 +860,6 @@ "displayName": "Qwen3 Coder Plus", "description": "Hosted Qwen coder for software agents, repo edits, and long-context code", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -920,7 +878,6 @@ "displayName": "Qwen3-LiveTranslate Flash Realtime", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 53248, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-04", @@ -931,7 +888,7 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -939,7 +896,6 @@ "displayName": "Qwen3 Max", "description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -958,7 +914,6 @@ "displayName": "Qwen3-Next 80B-A3B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -977,7 +932,6 @@ "displayName": "Qwen3-Next 80B-A3B (Thinking)", "description": "Efficient Qwen thinking model for local reasoning, math, and coding agents", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -996,7 +950,6 @@ "displayName": "Qwen3-Omni Flash", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 65536, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-04", @@ -1010,7 +963,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -1018,7 +971,6 @@ "displayName": "Qwen3-Omni Flash Realtime", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 65536, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-04", @@ -1029,7 +981,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -1037,7 +989,6 @@ "displayName": "Qwen3-VL 235B-A22B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -1056,7 +1007,6 @@ "displayName": "Qwen3-VL 30B-A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -1075,7 +1025,6 @@ "displayName": "Qwen3-VL Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -1097,7 +1046,6 @@ "displayName": "Qwen3.5 122B-A10B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -1111,7 +1059,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -1119,7 +1067,6 @@ "displayName": "Qwen3.5 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -1133,7 +1080,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -1141,7 +1088,6 @@ "displayName": "Qwen3.5 35B-A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -1155,7 +1101,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -1163,7 +1109,6 @@ "displayName": "Qwen3.5 397B-A17B", "description": "Large open Qwen multimodal MoE for visual agents and long technical tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -1177,7 +1122,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -1185,7 +1130,6 @@ "displayName": "Qwen3.5 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -1199,7 +1143,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -1207,7 +1151,6 @@ "displayName": "Qwen3.6 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -1221,7 +1164,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -1229,7 +1172,6 @@ "displayName": "Qwen3.6 35B-A3B", "description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -1243,7 +1185,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -1251,7 +1193,6 @@ "displayName": "Qwen3.6 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -1265,7 +1206,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -1273,7 +1214,6 @@ "displayName": "Qwen3.6 Max Preview", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -1295,7 +1235,6 @@ "displayName": "Qwen3.6 Plus", "description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -1309,7 +1248,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -1317,7 +1256,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "lastUpdated": "2026-05-21", @@ -1338,7 +1276,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -1352,7 +1289,29 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "qwen3.8-flash": { + "displayName": "Qwen3.8 Flash", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"], + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -1360,7 +1319,6 @@ "displayName": "Qwen3.8 Max", "description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -1375,7 +1333,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } }, @@ -1383,7 +1341,6 @@ "displayName": "QwQ Plus", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -1404,7 +1361,6 @@ "displayName": "DeepSeek R1", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2025-01-01", @@ -1422,7 +1378,6 @@ "displayName": "DeepSeek R1 0528", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2025-05-28", @@ -1440,7 +1395,6 @@ "displayName": "DeepSeek R1 Distill Llama 70B", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2025-01-01", @@ -1458,10 +1412,10 @@ "displayName": "DeepSeek R1 Distill Llama 8B", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2025-01-01", + "isFree": true, "capabilities": { "vision": false, "reasoning": true, @@ -1476,10 +1430,10 @@ "displayName": "DeepSeek R1 Distill Qwen 1.5B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2025-01-01", + "isFree": true, "capabilities": { "vision": false, "reasoning": true, @@ -1494,7 +1448,6 @@ "displayName": "DeepSeek R1 Distill Qwen 14B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2025-01-01", @@ -1512,7 +1465,6 @@ "displayName": "DeepSeek R1 Distill Qwen 32B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2025-01-01", @@ -1530,7 +1482,6 @@ "displayName": "DeepSeek R1 Distill Qwen 7B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2025-01-01", @@ -1548,7 +1499,6 @@ "displayName": "DeepSeek V3", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 65536, "maxOutputTokens": 8192, "lastUpdated": "2024-12-01", @@ -1566,7 +1516,6 @@ "displayName": "DeepSeek V3.1", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 65536, "lastUpdated": "2025-01-01", @@ -1584,7 +1533,6 @@ "displayName": "DeepSeek V3.2 Exp", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 65536, "lastUpdated": "2025-01-01", @@ -1602,7 +1550,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -1626,7 +1573,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -1650,7 +1596,6 @@ "displayName": "GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 202752, "maxOutputTokens": 16384, "lastUpdated": "2026-02-11", @@ -1671,7 +1616,6 @@ "displayName": "GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 202752, "maxOutputTokens": 128000, "structuredOutput": true, @@ -1693,7 +1637,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -1712,7 +1655,6 @@ "displayName": "Moonshot Kimi K2 Thinking", "description": "Kimi reasoning model for long-horizon research, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 16384, "structuredOutput": true, @@ -1731,7 +1673,6 @@ "displayName": "Moonshot Kimi K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -1746,7 +1687,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -1754,7 +1695,6 @@ "displayName": "Moonshot Kimi K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-01", @@ -1769,7 +1709,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -1777,7 +1717,6 @@ "displayName": "kimi/kimi-k2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -1792,7 +1731,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -1800,7 +1739,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-12", @@ -1818,7 +1756,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -1836,7 +1773,6 @@ "displayName": "Moonshot Kimi K2 Instruct", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "structuredOutput": false, @@ -1855,7 +1791,6 @@ "displayName": "QVQ Max", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -1874,7 +1809,6 @@ "displayName": "Qwen Deep Research", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -1893,7 +1827,6 @@ "displayName": "Qwen Doc Turbo", "description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -1912,7 +1845,6 @@ "displayName": "Qwen Flash", "description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -1934,7 +1866,6 @@ "displayName": "Qwen Long", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 10000000, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -1953,7 +1884,6 @@ "displayName": "Qwen Math Plus", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 4096, "maxOutputTokens": 3072, "knowledgeCutoff": "2024-04", @@ -1972,7 +1902,6 @@ "displayName": "Qwen Math Turbo", "description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 4096, "maxOutputTokens": 3072, "knowledgeCutoff": "2024-04", @@ -1991,7 +1920,6 @@ "displayName": "Qwen Max", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2010,7 +1938,6 @@ "displayName": "Qwen-MT Plus", "description": "Translation model for multilingual conversion, localization, and cross-language workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 16384, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2029,7 +1956,6 @@ "displayName": "Qwen-MT Turbo", "description": "Translation model for multilingual conversion, localization, and cross-language workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 16384, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2048,7 +1974,6 @@ "displayName": "Qwen-Omni Turbo", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-04", @@ -2059,7 +1984,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -2067,7 +1992,6 @@ "displayName": "Qwen-Omni Turbo Realtime", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-04", @@ -2086,7 +2010,6 @@ "displayName": "Qwen Plus", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -2108,7 +2031,6 @@ "displayName": "Qwen Plus Character", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-04", @@ -2127,7 +2049,6 @@ "displayName": "Qwen Turbo", "description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-04", @@ -2149,7 +2070,6 @@ "displayName": "Qwen-VL Max", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2168,7 +2088,6 @@ "displayName": "Qwen-VL OCR", "description": "OCR model for extracting structured text from documents and screenshots", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 34096, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-04", @@ -2187,7 +2106,6 @@ "displayName": "Qwen-VL Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2206,7 +2124,6 @@ "displayName": "Qwen2.5 14B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2225,7 +2142,6 @@ "displayName": "Qwen2.5 32B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2244,7 +2160,6 @@ "displayName": "Qwen2.5 72B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2263,7 +2178,6 @@ "displayName": "Qwen2.5 7B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2282,7 +2196,6 @@ "displayName": "Qwen2.5-Coder 32B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2301,7 +2214,6 @@ "displayName": "Qwen2.5-Coder 7B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2320,7 +2232,6 @@ "displayName": "Qwen2.5-Math 72B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 4096, "maxOutputTokens": 3072, "knowledgeCutoff": "2024-04", @@ -2339,7 +2250,6 @@ "displayName": "Qwen2.5-Math 7B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 4096, "maxOutputTokens": 3072, "knowledgeCutoff": "2024-04", @@ -2358,7 +2268,6 @@ "displayName": "Qwen2.5-Omni 7B", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 32768, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-04", @@ -2369,7 +2278,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -2377,7 +2286,6 @@ "displayName": "Qwen2.5-VL 72B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2396,7 +2304,6 @@ "displayName": "Qwen2.5-VL 7B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2415,7 +2322,6 @@ "displayName": "Qwen3 14B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-04", @@ -2437,7 +2343,6 @@ "displayName": "Qwen3 235B-A22B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -2459,7 +2364,6 @@ "displayName": "Qwen3 32B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -2481,7 +2385,6 @@ "displayName": "Qwen3 8B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-04", @@ -2503,7 +2406,6 @@ "displayName": "Qwen3-ASR Flash", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 53248, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-04", @@ -2522,7 +2424,6 @@ "displayName": "Qwen3-Coder 30B-A3B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2541,7 +2442,6 @@ "displayName": "Qwen3-Coder 480B-A35B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2560,7 +2460,6 @@ "displayName": "Qwen3 Coder Flash", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2579,7 +2478,6 @@ "displayName": "Qwen3 Coder Plus", "description": "Hosted Qwen coder for software agents, repo edits, and long-context code", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2598,7 +2496,6 @@ "displayName": "Qwen3 Max", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2617,7 +2514,6 @@ "displayName": "Qwen3-Next 80B-A3B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -2636,7 +2532,6 @@ "displayName": "Qwen3-Next 80B-A3B (Thinking)", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -2655,7 +2550,6 @@ "displayName": "Qwen3-Omni Flash", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 65536, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-04", @@ -2669,7 +2563,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text", "audio"] } }, @@ -2677,7 +2571,6 @@ "displayName": "Qwen3-Omni Flash Realtime", "description": "Qwen omni model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 65536, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-04", @@ -2696,7 +2589,6 @@ "displayName": "Qwen3-VL 235B-A22B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -2715,7 +2607,6 @@ "displayName": "Qwen3-VL 30B-A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -2734,7 +2625,6 @@ "displayName": "Qwen3-VL Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -2756,7 +2646,6 @@ "displayName": "Qwen3.5 397B-A17B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2770,7 +2659,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -2778,7 +2667,6 @@ "displayName": "Qwen3.5 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2793,7 +2681,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -2801,7 +2689,6 @@ "displayName": "Qwen3.5 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2815,7 +2702,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -2823,7 +2710,6 @@ "displayName": "Qwen3.6 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -2837,7 +2723,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -2845,7 +2731,6 @@ "displayName": "Qwen3.6 Max Preview", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 245800, "maxOutputTokens": 65536, "structuredOutput": true, @@ -2867,7 +2752,6 @@ "displayName": "Qwen3.6 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -2881,7 +2765,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -2889,7 +2773,6 @@ "displayName": "Qwen3.7 Flash", "description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "inputLimit": 991000, "maxOutputTokens": 65536, @@ -2904,7 +2787,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -2912,7 +2795,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "lastUpdated": "2026-05-21", @@ -2933,7 +2815,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04", @@ -2947,7 +2828,29 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "qwen3.8-flash": { + "displayName": "Qwen3.8 Flash", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"], + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -2955,7 +2858,6 @@ "displayName": "Qwen3.8 Max", "description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -2970,7 +2872,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } }, @@ -2978,7 +2880,6 @@ "displayName": "QwQ 32B", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -2997,7 +2898,6 @@ "displayName": "QwQ Plus", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04", @@ -3016,7 +2916,6 @@ "displayName": "siliconflow/deepseek-r1-0528", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 163840, "maxOutputTokens": 32768, "structuredOutput": true, @@ -3035,7 +2934,6 @@ "displayName": "siliconflow/deepseek-v3-0324", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 163840, "maxOutputTokens": 163840, "structuredOutput": true, @@ -3054,7 +2952,6 @@ "displayName": "siliconflow/deepseek-v3.1-terminus", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 163840, "maxOutputTokens": 65536, "structuredOutput": true, @@ -3076,7 +2973,6 @@ "displayName": "siliconflow/deepseek-v3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 163840, "maxOutputTokens": 65536, "structuredOutput": true, @@ -3098,7 +2994,6 @@ "displayName": "Tongyi Intent Detect V3", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/models", "contextWindow": 8192, "maxOutputTokens": 1024, "knowledgeCutoff": "2024-04", @@ -3119,7 +3014,6 @@ "displayName": "GLM-4.7", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 202752, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -3142,7 +3036,6 @@ "displayName": "GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 202752, "maxOutputTokens": 16384, "lastUpdated": "2026-02-11", @@ -3164,7 +3057,6 @@ "displayName": "Kimi K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -3187,7 +3079,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 196608, "maxOutputTokens": 24576, "lastUpdated": "2026-02-12", @@ -3206,7 +3097,6 @@ "displayName": "Qwen3 Coder Next", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -3226,7 +3116,6 @@ "displayName": "Qwen3 Coder Plus", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -3246,7 +3135,6 @@ "displayName": "Qwen3 Max", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -3266,7 +3154,6 @@ "displayName": "Qwen3.5 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -3289,7 +3176,6 @@ "displayName": "Qwen3.6 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -3303,7 +3189,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3311,7 +3197,6 @@ "displayName": "Qwen3.6 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -3326,7 +3211,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3334,7 +3219,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "lastUpdated": "2026-05-21", @@ -3355,7 +3239,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://help.aliyun.com/zh/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04", @@ -3370,7 +3253,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -3380,7 +3263,6 @@ "displayName": "GLM-4.7", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 202752, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -3403,7 +3285,6 @@ "displayName": "GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 202752, "maxOutputTokens": 16384, "lastUpdated": "2026-02-11", @@ -3425,7 +3306,6 @@ "displayName": "Kimi K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -3440,7 +3320,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3448,7 +3328,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 196608, "inputLimit": 196601, "maxOutputTokens": 24576, @@ -3468,7 +3347,6 @@ "displayName": "Qwen3 Coder Next", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -3488,7 +3366,6 @@ "displayName": "Qwen3 Coder Plus", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -3508,7 +3385,6 @@ "displayName": "Qwen3 Max", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -3528,7 +3404,6 @@ "displayName": "Qwen3.5 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -3543,7 +3418,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3551,7 +3426,6 @@ "displayName": "Qwen3.6 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -3565,7 +3439,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3573,7 +3447,6 @@ "displayName": "Qwen3.6 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -3588,7 +3461,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3596,7 +3469,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 65536, "lastUpdated": "2026-05-21", @@ -3617,7 +3489,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/coding-plan", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04", @@ -3632,7 +3503,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -3642,7 +3513,6 @@ "displayName": "DeepSeek V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 131072, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -3666,7 +3536,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -3691,7 +3560,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -3716,7 +3584,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -3741,11 +3608,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "isFree": true, "capabilities": { "vision": false, @@ -3765,7 +3631,6 @@ "displayName": "GLM-5", "description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 202752, "maxOutputTokens": 16384, "structuredOutput": true, @@ -3788,7 +3653,6 @@ "displayName": "GLM-5.1", "description": "Strong GLM coding model for agentic engineering, terminals, and repository generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 202752, "maxOutputTokens": 128000, "structuredOutput": true, @@ -3811,7 +3675,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -3834,7 +3697,6 @@ "displayName": "HappyHorse 1.1 Image-to-Video", "description": "Video model for image-to-video generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-17", @@ -3846,14 +3708,13 @@ }, "modalities": { "input": ["image", "text"], - "output": [] + "output": ["video"] } }, "happyhorse-1.1-r2v": { "displayName": "HappyHorse 1.1 Reference-to-Video", "description": "Video model for reference-guided video generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-17", @@ -3865,14 +3726,13 @@ }, "modalities": { "input": ["image", "text"], - "output": [] + "output": ["video"] } }, "happyhorse-1.1-t2v": { "displayName": "HappyHorse 1.1 Text-to-Video", "description": "Video model for prompt-driven text-to-video generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-17", @@ -3884,14 +3744,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "kimi-k2.5": { "displayName": "Kimi K2.5", "description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 262144, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-01", @@ -3907,7 +3766,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3915,7 +3774,6 @@ "displayName": "Kimi K2.6", "description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -3931,7 +3789,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3939,7 +3797,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -3955,7 +3812,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -3963,7 +3820,6 @@ "displayName": "MiniMax-M2.5", "description": "Prior MiniMax coding model for agent workflows, office edits, and automation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 196608, "inputLimit": 196601, "maxOutputTokens": 32768, @@ -3983,7 +3839,6 @@ "displayName": "Qwen Image 2.0", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-03-03", @@ -4002,7 +3857,6 @@ "displayName": "Qwen Image 2.0 Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-03-03", @@ -4021,7 +3875,6 @@ "displayName": "Qwen3.6 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -4036,7 +3889,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4044,7 +3897,6 @@ "displayName": "Qwen3.6 Plus", "description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -4060,7 +3912,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4068,7 +3920,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -4091,7 +3942,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -4107,7 +3957,30 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "qwen3.8-flash": { + "displayName": "Qwen3.8 Flash", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "isFree": true, + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"], + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4115,7 +3988,6 @@ "displayName": "Qwen3.8 Max", "description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -4131,7 +4003,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } }, @@ -4139,7 +4011,6 @@ "displayName": "Qwen3.8 Max Preview", "description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows", "lifecycle": "beta", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -4154,7 +4025,7 @@ "efforts": ["low", "medium", "xhigh"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4162,7 +4033,6 @@ "displayName": "Wan2.7 Image", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-05-29", @@ -4181,7 +4051,6 @@ "displayName": "Wan2.7 Image Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/zh/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-05-29", @@ -4202,7 +4071,6 @@ "displayName": "DeepSeek V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 131072, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -4226,7 +4094,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -4251,7 +4118,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -4276,7 +4142,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -4301,11 +4166,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "isFree": true, "capabilities": { "vision": false, @@ -4325,7 +4189,6 @@ "displayName": "GLM-5", "description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 202752, "maxOutputTokens": 16384, "structuredOutput": true, @@ -4348,7 +4211,6 @@ "displayName": "GLM-5.1", "description": "Strong GLM coding model for agentic engineering, terminals, and repository generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 202752, "maxOutputTokens": 128000, "structuredOutput": true, @@ -4371,7 +4233,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -4394,7 +4255,6 @@ "displayName": "HappyHorse 1.1 Image-to-Video", "description": "Video model for image-to-video generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-17", @@ -4406,14 +4266,13 @@ }, "modalities": { "input": ["image", "text"], - "output": [] + "output": ["video"] } }, "happyhorse-1.1-r2v": { "displayName": "HappyHorse 1.1 Reference-to-Video", "description": "Video model for reference-guided video generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-17", @@ -4425,14 +4284,13 @@ }, "modalities": { "input": ["image", "text"], - "output": [] + "output": ["video"] } }, "happyhorse-1.1-t2v": { "displayName": "HappyHorse 1.1 Text-to-Video", "description": "Video model for prompt-driven text-to-video generation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-17", @@ -4444,14 +4302,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "kimi-k2.5": { "displayName": "Kimi K2.5", "description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 262144, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-01", @@ -4467,7 +4324,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4475,7 +4332,6 @@ "displayName": "Kimi K2.6", "description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -4491,7 +4347,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4499,7 +4355,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -4515,7 +4370,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4523,7 +4378,6 @@ "displayName": "MiniMax-M2.5", "description": "Prior MiniMax coding model for agent workflows, office edits, and automation", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 196608, "inputLimit": 196601, "maxOutputTokens": 32768, @@ -4543,7 +4397,6 @@ "displayName": "Qwen Image 2.0", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-03-03", @@ -4562,7 +4415,6 @@ "displayName": "Qwen Image 2.0 Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-03-03", @@ -4581,7 +4433,6 @@ "displayName": "Qwen3.6 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -4596,7 +4447,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4604,7 +4455,6 @@ "displayName": "Qwen3.6 Plus", "description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -4620,7 +4470,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4628,7 +4478,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -4651,7 +4500,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -4667,7 +4515,30 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "qwen3.8-flash": { + "displayName": "Qwen3.8 Flash", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "isFree": true, + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"], + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4675,7 +4546,6 @@ "displayName": "Qwen3.8 Max", "description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -4691,7 +4561,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } }, @@ -4699,7 +4569,6 @@ "displayName": "Qwen3.8 Max Preview", "description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows", "lifecycle": "beta", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -4714,7 +4583,7 @@ "efforts": ["low", "medium", "xhigh"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -4722,7 +4591,6 @@ "displayName": "Wan2.7 Image", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-05-29", @@ -4741,7 +4609,6 @@ "displayName": "Wan2.7 Image Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview", "contextWindow": 8192, "maxOutputTokens": 0, "lastUpdated": "2026-05-29", @@ -4762,7 +4629,6 @@ "displayName": "Gemma 4 31B IT", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "beta", - "docsUrl": "https://inference-docs.cerebras.ai/models/overview", "contextWindow": 131072, "maxOutputTokens": 40960, "structuredOutput": true, @@ -4784,7 +4650,6 @@ "displayName": "GPT OSS 120B", "description": "Open GPT reasoning model for self-hosted agents and controllable deployments", "lifecycle": "active", - "docsUrl": "https://inference-docs.cerebras.ai/models/overview", "contextWindow": 131072, "maxOutputTokens": 40960, "structuredOutput": true, @@ -4808,7 +4673,6 @@ "displayName": "Aya Expanse 32B", "description": "Open multilingual model optimized for generation across 23 languages", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 4000, "lastUpdated": "2024-10-24", @@ -4826,7 +4690,6 @@ "displayName": "Aya Expanse 8B", "description": "Compact open multilingual model optimized for generation across 23 languages", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 8000, "maxOutputTokens": 4000, "lastUpdated": "2024-10-24", @@ -4844,7 +4707,6 @@ "displayName": "Aya Vision 32B", "description": "Open multilingual vision model for OCR, visual reasoning, and image question answering", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 16000, "maxOutputTokens": 4000, "lastUpdated": "2025-05-14", @@ -4862,7 +4724,6 @@ "displayName": "Aya Vision 8B", "description": "Compact open multilingual vision model for OCR and visual question answering", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 16000, "maxOutputTokens": 4000, "lastUpdated": "2025-05-14", @@ -4880,7 +4741,6 @@ "displayName": "Command A", "description": "Cohere command model for multilingual enterprise agents, tools, and chat", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 256000, "maxOutputTokens": 8000, "knowledgeCutoff": "2024-06-01", @@ -4899,7 +4759,6 @@ "displayName": "Command A Plus", "description": "Cohere's stronger command model for multilingual agents and enterprise workflows", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04-01", @@ -4922,7 +4781,6 @@ "displayName": "Command A Reasoning", "description": "Cohere reasoning model for multilingual enterprise agents, tools, and complex workflows", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 256000, "maxOutputTokens": 32000, "knowledgeCutoff": "2024-06-01", @@ -4944,7 +4802,6 @@ "displayName": "Command A Translate", "description": "Translation model for multilingual conversion, localization, and cross-language workflows", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 8000, "maxOutputTokens": 8000, "knowledgeCutoff": "2024-06-01", @@ -4963,7 +4820,6 @@ "displayName": "Command A Vision", "description": "Cohere vision model for multilingual document analysis, OCR, and image understanding", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 8000, "knowledgeCutoff": "2024-06-01", @@ -4982,7 +4838,6 @@ "displayName": "Command R", "description": "Cohere retrieval model for long-context chat and enterprise RAG workflows", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 4000, "knowledgeCutoff": "2024-06-01", @@ -5001,7 +4856,6 @@ "displayName": "Command R+", "description": "Cohere's RAG workhorse for long-context enterprise search and tool use", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 4000, "knowledgeCutoff": "2024-06-01", @@ -5020,7 +4874,6 @@ "displayName": "Command R7B", "description": "Cohere retrieval model for long-context chat and enterprise RAG workflows", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 4000, "knowledgeCutoff": "2024-06-01", @@ -5039,7 +4892,6 @@ "displayName": "Command R7B Arabic", "description": "Open Command R model optimized for Arabic enterprise chat, RAG, and cultural knowledge", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 4000, "knowledgeCutoff": "2024-06-01", @@ -5058,7 +4910,6 @@ "displayName": "North Mini Code", "description": "Cohere coding model for practical software engineering and agentic edits", "lifecycle": "active", - "docsUrl": "https://docs.cohere.com/docs/models", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-09-23", @@ -5084,7 +4935,6 @@ "displayName": "Gemma Sea Lion V4 27B It", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "structuredOutput": false, @@ -5103,7 +4953,6 @@ "displayName": "Deepseek R1 Distill Qwen 32B", "description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 80000, "maxOutputTokens": 80000, "knowledgeCutoff": "2024-07", @@ -5123,7 +4972,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 1310720, "maxOutputTokens": 1048576, "knowledgeCutoff": "2025-05", @@ -5147,11 +4995,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 1048576, "maxOutputTokens": 1048576, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -5170,7 +5017,6 @@ "displayName": "Gemma 4 26B A4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 256000, "maxOutputTokens": 16384, "structuredOutput": true, @@ -5193,7 +5039,6 @@ "displayName": "Granite 4.0 H Micro", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": false, @@ -5212,7 +5057,6 @@ "displayName": "Llama 3.1 8B Instruct fp8", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 32000, "maxOutputTokens": 32000, "knowledgeCutoff": "2023-12", @@ -5232,7 +5076,6 @@ "displayName": "Llama 3.2 11B Vision Instruct", "description": "Open Llama multimodal model for image understanding and text reasoning", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2023-12", @@ -5252,7 +5095,6 @@ "displayName": "Llama 3.2 1B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 60000, "maxOutputTokens": 60000, "knowledgeCutoff": "2023-12", @@ -5272,7 +5114,6 @@ "displayName": "Llama 3.2 3B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 80000, "maxOutputTokens": 80000, "knowledgeCutoff": "2023-12", @@ -5292,7 +5133,6 @@ "displayName": "Llama 3.3 70B Instruct fp8 Fast", "description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 24000, "maxOutputTokens": 24000, "knowledgeCutoff": "2023-12", @@ -5312,7 +5152,6 @@ "displayName": "Llama 4 Scout 17B 16E Instruct", "description": "Open Llama with long-context vision for efficient multimodal agents", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 131000, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-08", @@ -5332,7 +5171,6 @@ "displayName": "Llama Guard 3 8B", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2023-12", @@ -5352,7 +5190,6 @@ "displayName": "Mistral Small 3.1 24B Instruct", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "structuredOutput": false, @@ -5371,7 +5208,6 @@ "displayName": "Kimi K2.6", "description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 262144, "maxOutputTokens": 256000, "knowledgeCutoff": "2025-01", @@ -5395,7 +5231,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -5419,7 +5254,6 @@ "displayName": "Nemotron 3 Super 120B", "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 256000, "maxOutputTokens": 256000, "structuredOutput": true, @@ -5442,7 +5276,6 @@ "displayName": "GPT OSS 120B", "description": "Open GPT reasoning model for self-hosted agents and controllable deployments", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 128000, "maxOutputTokens": 16384, "structuredOutput": true, @@ -5464,7 +5297,6 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 128000, "maxOutputTokens": 16384, "structuredOutput": true, @@ -5483,7 +5315,6 @@ "displayName": "Qwen2.5 Coder 32B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 32768, "maxOutputTokens": 32768, "structuredOutput": false, @@ -5502,7 +5333,6 @@ "displayName": "Qwen3 30B A3b fp8", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 32768, "maxOutputTokens": 32768, "structuredOutput": false, @@ -5521,7 +5351,6 @@ "displayName": "Qwen3.8 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -5544,7 +5373,6 @@ "displayName": "Qwq 32B", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 24000, "maxOutputTokens": 24000, "knowledgeCutoff": "2024-04", @@ -5564,7 +5392,6 @@ "displayName": "GLM-4.7-Flash", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -5588,7 +5415,6 @@ "displayName": "Glm 5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://developers.cloudflare.com/workers-ai/models/", "contextWindow": 262144, "maxOutputTokens": 256000, "structuredOutput": true, @@ -5606,6 +5432,46 @@ "input": ["text"], "output": ["text"] } + }, + "@cf/zai-org/glm-5.3": { + "displayName": "Glm 5.3", + "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", + "lifecycle": "active", + "contextWindow": 1310720, + "maxOutputTokens": 1310720, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "high"], + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "@cf/zai-org/glm-5.3-flash": { + "displayName": "Glm 5.3 Flash", + "description": "GLM vision model for visual reasoning, documents, and multimodal agents", + "lifecycle": "active", + "contextWindow": 1310720, + "maxOutputTokens": 1048576, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } } }, "deepinfra": { @@ -5613,7 +5479,6 @@ "displayName": "Seed 2.0 Code", "description": "ByteDance Seed coding model for multimodal software engineering and long-running agents", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 256000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -5635,7 +5500,6 @@ "displayName": "Seed 2.0 Mini", "description": "Lightweight ByteDance Seed 2.0 model for low-latency multimodal reasoning and high-volume tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 256000, "maxOutputTokens": 32000, "structuredOutput": true, @@ -5654,7 +5518,6 @@ "displayName": "Seed 2.0 Pro", "description": "Flagship ByteDance Seed 2.0 model for complex multimodal reasoning and long-horizon agent workflows", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 256000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -5673,7 +5536,6 @@ "displayName": "DeepSeek-R1-0528", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 163840, "maxOutputTokens": 64000, "knowledgeCutoff": "2024-07", @@ -5693,7 +5555,6 @@ "displayName": "DeepSeek-V3", "description": "Open DeepSeek MoE chat model for coding, math, and general reasoning", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 163840, "maxOutputTokens": 8192, "structuredOutput": true, @@ -5712,7 +5573,6 @@ "displayName": "DeepSeek V3 0324", "description": "March 2025 checkpoint of DeepSeek-V3 with improved reasoning and coding", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 163840, "maxOutputTokens": 163840, "structuredOutput": true, @@ -5731,7 +5591,6 @@ "displayName": "DeepSeek-V3.1", "description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 163840, "maxOutputTokens": 8192, "structuredOutput": true, @@ -5753,7 +5612,6 @@ "displayName": "DeepSeek-V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 163840, "maxOutputTokens": 64000, "knowledgeCutoff": "2024-12", @@ -5776,7 +5634,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-05", @@ -5800,7 +5657,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -5823,7 +5679,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-05", @@ -5847,11 +5702,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -5870,7 +5724,6 @@ "displayName": "Gemma 4 26B A4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -5892,7 +5745,6 @@ "displayName": "Gemma 4 31B IT", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -5906,7 +5758,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -5914,7 +5766,6 @@ "displayName": "Gemma 4 E4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 131072, "maxOutputTokens": 8192, "structuredOutput": true, @@ -5936,7 +5787,6 @@ "displayName": "Llama 3.3 70B Turbo", "description": "Compact Llama instruction model for fast chat and local deployment", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 131072, "maxOutputTokens": 16384, "structuredOutput": true, @@ -5955,7 +5805,6 @@ "displayName": "Llama 4 Maverick 17B FP8", "description": "Open multimodal Llama model for strong reasoning and fast responses", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 16384, "structuredOutput": true, @@ -5974,7 +5823,6 @@ "displayName": "Llama 4 Scout 17B", "description": "Open multimodal Llama model for long-context analysis and efficient agents", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 327680, "maxOutputTokens": 16384, "structuredOutput": true, @@ -5993,7 +5841,6 @@ "displayName": "MiniMax M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "deprecated", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 196608, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-06", @@ -6012,7 +5859,6 @@ "displayName": "MiniMax-M2.7", "description": "Open MiniMax flagship for coding agents, office automation, and complex environments", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 196608, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -6030,9 +5876,8 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 524288, - "maxOutputTokens": 128000, + "maxOutputTokens": 512000, "structuredOutput": true, "lastUpdated": "2026-06-01", "capabilities": { @@ -6041,7 +5886,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6049,7 +5894,6 @@ "displayName": "Kimi K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -6064,7 +5908,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6072,7 +5916,6 @@ "displayName": "Kimi K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-04", @@ -6087,7 +5930,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6095,7 +5938,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -6110,7 +5952,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6118,7 +5960,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -6140,7 +5981,6 @@ "displayName": "Llama 3.3 Nemotron Super 49B v1.5", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "deprecated", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 131072, "maxOutputTokens": 131072, "structuredOutput": true, @@ -6159,7 +5999,6 @@ "displayName": "Nemotron 3 Nano 30B A3B", "description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "lastUpdated": "2025-12-15", @@ -6180,7 +6019,6 @@ "displayName": "Nemotron 3 Nano Omni 30B A3B Reasoning", "description": "Open Nemotron omni model combining reasoning with text, vision, and audio", "lifecycle": "deprecated", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -6191,7 +6029,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -6199,7 +6037,6 @@ "displayName": "GPT OSS 120B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 131072, "maxOutputTokens": 16384, "structuredOutput": true, @@ -6221,7 +6058,6 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 131072, "maxOutputTokens": 16384, "structuredOutput": true, @@ -6243,7 +6079,6 @@ "displayName": "Qwen3 235B-A22B Instruct 2507", "description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 16384, "structuredOutput": true, @@ -6262,7 +6097,6 @@ "displayName": "Qwen3 30B A3B", "description": "Sparse MoE Qwen model with 3B active parameters for efficient chat and reasoning", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 40960, "maxOutputTokens": 16384, "structuredOutput": true, @@ -6281,7 +6115,6 @@ "displayName": "Qwen3 32B", "description": "Dense open Qwen model for self-hosted chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 40960, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -6301,7 +6134,6 @@ "displayName": "Qwen3 Coder 480B A35B Instruct Turbo", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 66536, "knowledgeCutoff": "2025-04", @@ -6321,7 +6153,6 @@ "displayName": "Qwen3 Max", "description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 256000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -6341,7 +6172,6 @@ "displayName": "Qwen3-Next 80B-A3B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -6361,7 +6191,6 @@ "displayName": "Qwen3 VL 235B A22B Instruct", "description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -6381,7 +6210,6 @@ "displayName": "Qwen3.5 122B-A10B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -6392,7 +6220,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -6400,7 +6228,6 @@ "displayName": "Qwen3.5 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -6411,7 +6238,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -6419,7 +6246,6 @@ "displayName": "Qwen 3.5 35B A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 81920, "knowledgeCutoff": "2025-01", @@ -6431,7 +6257,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6439,7 +6265,6 @@ "displayName": "Qwen 3.5 397B A17B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 81920, "knowledgeCutoff": "2025-01", @@ -6451,7 +6276,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6459,7 +6284,6 @@ "displayName": "Qwen3.5 9B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -6470,7 +6294,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6478,7 +6302,6 @@ "displayName": "Qwen3.6 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -6489,7 +6312,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -6497,7 +6320,6 @@ "displayName": "Qwen3.6 35B A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 81920, "structuredOutput": true, @@ -6508,7 +6330,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6516,7 +6338,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 256000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -6535,7 +6356,6 @@ "displayName": "Qwen3.8 2.4T A95B", "description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -6557,7 +6377,6 @@ "displayName": "Qwen3.8 27B", "description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -6580,7 +6399,6 @@ "displayName": "Qwen3.8 Max", "description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 256000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -6591,7 +6409,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } }, @@ -6599,7 +6417,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 256000, "knowledgeCutoff": "2026-03-01", @@ -6610,7 +6427,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -6618,7 +6435,6 @@ "displayName": "Hy3", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 64000, "structuredOutput": true, @@ -6637,7 +6453,6 @@ "displayName": "Inkling", "description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 524288, "maxOutputTokens": 1048576, "lastUpdated": "2026-07-15", @@ -6655,7 +6470,6 @@ "displayName": "Inkling Small", "description": "Multimodal MoE reasoning model (276B total, 12B active) for text, image, and audio", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 524288, "maxOutputTokens": 1048576, "structuredOutput": true, @@ -6674,7 +6488,6 @@ "displayName": "MiMo-V2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 262144, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-12", @@ -6689,7 +6502,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -6697,7 +6510,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-12", @@ -6720,7 +6532,6 @@ "displayName": "GLM-4.6", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 202752, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -6743,7 +6554,6 @@ "displayName": "GLM-4.7", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 202752, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -6766,7 +6576,6 @@ "displayName": "GLM-4.7-Flash", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 202752, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -6786,7 +6595,6 @@ "displayName": "GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 202752, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-12", @@ -6809,7 +6617,6 @@ "displayName": "GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 202752, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -6832,7 +6639,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://deepinfra.com/models", "contextWindow": 1048576, "maxOutputTokens": 32768, "structuredOutput": true, @@ -6850,125 +6656,55 @@ "input": ["text"], "output": ["text"] } - } - }, - "deepseek": { - "deepseek-chat": { - "displayName": "DeepSeek Chat", - "description": "DeepSeek chat model for instruction following, coding, and analysis", - "lifecycle": "active", - "docsUrl": "https://api-docs.deepseek.com/quick_start/pricing", - "contextWindow": 1000000, - "maxOutputTokens": 384000, - "knowledgeCutoff": "2025-09", - "lastUpdated": "2026-02-28", - "capabilities": { - "vision": false, - "reasoning": false, - "functionCalling": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, - "deepseek-reasoner": { - "displayName": "DeepSeek Reasoner", - "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", - "lifecycle": "active", - "docsUrl": "https://api-docs.deepseek.com/quick_start/pricing", - "contextWindow": 1000000, - "maxOutputTokens": 384000, - "knowledgeCutoff": "2025-09", - "lastUpdated": "2026-02-28", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } }, - "deepseek-v4-flash": { - "displayName": "DeepSeek V4 Flash", - "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", + "zai-org/GLM-5.3": { + "displayName": "GLM-5.3", + "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", "lifecycle": "active", - "docsUrl": "https://api-docs.deepseek.com/quick_start/pricing", - "contextWindow": 1000000, - "maxOutputTokens": 384000, - "knowledgeCutoff": "2025-05", + "contextWindow": 1048576, + "maxOutputTokens": 131072, "structuredOutput": true, - "lastUpdated": "2026-07-31", + "lastUpdated": "2026-08-14", "capabilities": { "vision": false, "reasoning": true, "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "high", "max"], - "toggle": true + "efforts": ["low", "high", "max"] }, "modalities": { "input": ["text"], "output": ["text"] } }, - "deepseek-v4-pro": { - "displayName": "DeepSeek V4 Pro", - "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", + "zai-org/GLM-5.3-Flash": { + "displayName": "GLM-5.3-Flash", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", "lifecycle": "active", - "docsUrl": "https://api-docs.deepseek.com/quick_start/pricing", - "contextWindow": 1000000, - "maxOutputTokens": 384000, + "contextWindow": 1048576, + "maxOutputTokens": 131072, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-26", "capabilities": { - "vision": false, + "vision": true, "reasoning": true, "functionCalling": true }, "thinkingOptions": { - "efforts": ["high", "max"], - "toggle": true + "efforts": ["low", "high", "max"] }, "modalities": { - "input": ["text"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } } }, - "fireworks-ai": { - "accounts/fireworks/models/deepseek-v4-flash": { + "deepseek": { + "deepseek-v4-flash": { "displayName": "DeepSeek V4 Flash", - "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", - "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", - "contextWindow": 1000000, - "maxOutputTokens": 384000, - "knowledgeCutoff": "2025-05", - "structuredOutput": true, - "lastUpdated": "2026-06-16", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "efforts": ["high", "max"], - "toggle": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, - "accounts/fireworks/models/deepseek-v4-flash-0731": { - "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -6988,16 +6724,36 @@ "output": ["text"] } }, - "accounts/fireworks/models/deepseek-v4-pro": { + "deepseek-v4-flash-vision-exp": { + "displayName": "DeepSeek V4 Flash Vision Exp", + "description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work", + "lifecycle": "beta", + "contextWindow": 1000000, + "maxOutputTokens": 384000, + "structuredOutput": true, + "lastUpdated": "2026-08-21", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"], + "toggle": true + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, + "deepseek-v4-pro": { "displayName": "DeepSeek V4 Pro", - "description": "Open MoE flagship with million-token context for coding and long agent runs", + "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1000000, "maxOutputTokens": 384000, - "knowledgeCutoff": "2025-05", "structuredOutput": true, - "lastUpdated": "2026-04-24", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -7011,16 +6767,40 @@ "input": ["text"], "output": ["text"] } + } + }, + "fireworks-ai": { + "accounts/fireworks/models/deepseek-v4-flash-0731": { + "displayName": "DeepSeek V4 Flash 0731", + "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 384000, + "knowledgeCutoff": "2025-05", + "structuredOutput": true, + "lastUpdated": "2026-07-31", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"], + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } }, "accounts/fireworks/models/deepseek-v4-pro-0813": { "displayName": "DeepSeek V4 Pro 0813", "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1000000, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -7039,7 +6819,6 @@ "displayName": "GLM 5.2", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1048575, "maxOutputTokens": 131072, "lastUpdated": "2026-06-16", @@ -7057,35 +6836,55 @@ "output": ["text"] } }, - "accounts/fireworks/models/gpt-oss-120b": { - "displayName": "GPT OSS 120B", - "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", + "accounts/fireworks/models/glm-5p3": { + "displayName": "GLM 5.3", + "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", - "contextWindow": 131072, - "maxOutputTokens": 32768, - "lastUpdated": "2026-06-16", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-28", "capabilities": { "vision": false, "reasoning": true, "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"] + "efforts": ["high", "max"] }, "modalities": { "input": ["text"], "output": ["text"] } }, - "accounts/fireworks/models/gpt-oss-20b": { - "displayName": "GPT OSS 20B", + "accounts/fireworks/models/glm-5p3-flash": { + "displayName": "GLM 5.3 Flash", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["high", "max"] + }, + "modalities": { + "input": ["text", "image", "video", "pdf"], + "output": ["text"] + } + }, + "accounts/fireworks/models/gpt-oss-120b": { + "displayName": "GPT OSS 120B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 131072, "maxOutputTokens": 32768, - "lastUpdated": "2025-08-05", + "lastUpdated": "2026-06-16", "capabilities": { "vision": false, "reasoning": true, @@ -7103,7 +6902,6 @@ "displayName": "Inkling", "description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1048576, "maxOutputTokens": 1048576, "lastUpdated": "2026-07-15", @@ -7121,7 +6919,6 @@ "displayName": "Kimi K2.6", "description": "Kimi reasoning model for long-horizon research, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 262000, "maxOutputTokens": 262000, "lastUpdated": "2026-04-17", @@ -7142,7 +6939,6 @@ "displayName": "Kimi K2.7 Code", "description": "Kimi coding model for software agents, refactors, and repository reasoning", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 262000, "maxOutputTokens": 262000, "lastUpdated": "2026-06-16", @@ -7163,7 +6959,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -7182,32 +6977,10 @@ "output": ["text"] } }, - "accounts/fireworks/models/minimax-m2p7": { - "displayName": "MiniMax-M2.7", - "description": "MiniMax model for chat, coding, office work, and agentic tasks", - "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", - "contextWindow": 196608, - "maxOutputTokens": 196608, - "lastUpdated": "2026-04-12", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "efforts": ["low", "medium", "high"] - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "accounts/fireworks/models/minimax-m3": { "displayName": "MiniMax-M3", "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 512000, "maxOutputTokens": 512000, "lastUpdated": "2026-06-12", @@ -7220,7 +6993,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -7228,7 +7001,6 @@ "displayName": "Muse Glimmer 30B", "description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2026-01-04", @@ -7251,7 +7023,6 @@ "displayName": "Nemotron 3 Ultra 550B A55B", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 262144, "maxOutputTokens": 128000, "lastUpdated": "2026-06-04", @@ -7272,7 +7043,6 @@ "displayName": "Nemotron 3.5 Lightning 30B A3B", "description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -7294,7 +7064,6 @@ "displayName": "Qwen 3.7 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 262144, "maxOutputTokens": 65536, "lastUpdated": "2026-06-12", @@ -7316,7 +7085,6 @@ "displayName": "Qwen3.8 Max", "description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 262144, "maxOutputTokens": 131072, "lastUpdated": "2026-08-03", @@ -7337,7 +7105,6 @@ "displayName": "GLM 5.2 Fast", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1048575, "maxOutputTokens": 131072, "lastUpdated": "2026-06-26", @@ -7355,74 +7122,10 @@ "output": ["text"] } }, - "accounts/fireworks/routers/kimi-k2p6-fast": { - "displayName": "Kimi K2.6 Fast", - "description": "Kimi reasoning model for long-horizon research, planning, and tool use", - "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", - "contextWindow": 262000, - "maxOutputTokens": 262000, - "lastUpdated": "2026-06-05", - "capabilities": { - "vision": true, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "toggle": true - }, - "modalities": { - "input": ["text", "image"], - "output": ["text"] - } - }, - "accounts/fireworks/routers/kimi-k2p6-turbo": { - "displayName": "Kimi K2.6 Turbo", - "description": "Kimi reasoning model for long-horizon research, planning, and tool use", - "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", - "contextWindow": 262000, - "maxOutputTokens": 262000, - "lastUpdated": "2026-04-17", - "capabilities": { - "vision": true, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "toggle": true - }, - "modalities": { - "input": ["text", "image"], - "output": ["text"] - } - }, - "accounts/fireworks/routers/kimi-k2p7-code-fast": { - "displayName": "Kimi K2.7 Code Fast", - "description": "Kimi coding model for software agents, refactors, and repository reasoning", - "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", - "contextWindow": 262000, - "maxOutputTokens": 262000, - "lastUpdated": "2026-06-16", - "capabilities": { - "vision": true, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "toggle": true - }, - "modalities": { - "input": ["text", "image"], - "output": ["text"] - } - }, "accounts/fireworks/routers/kimi-k3-fast": { "displayName": "Kimi K3 Fast", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://fireworks.ai/docs/", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -7447,7 +7150,6 @@ "displayName": "Claude Fable 5", "description": "Claude model for creative writing, analysis, and controlled agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -7469,7 +7171,6 @@ "displayName": "Claude Haiku 4.5 (latest)", "description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 136000, "maxOutputTokens": 64000, @@ -7489,7 +7190,6 @@ "displayName": "Claude Opus 4.5 (latest)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 168000, "maxOutputTokens": 32000, @@ -7509,7 +7209,6 @@ "displayName": "Claude Opus 4.6", "description": "High-end Claude for difficult coding, planning, and slower expert reasoning", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 168000, "maxOutputTokens": 32000, @@ -7532,7 +7231,6 @@ "displayName": "Claude Opus 4.7", "description": "Stronger Opus tier for advanced software work and high-stakes reasoning", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 168000, "maxOutputTokens": 32000, @@ -7555,7 +7253,6 @@ "displayName": "Claude Opus 4.8", "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 168000, "maxOutputTokens": 64000, @@ -7578,7 +7275,6 @@ "displayName": "Claude Opus 5", "description": "Strongest Claude Opus model for coding, agents, and professional work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1000000, "inputLimit": 936000, "maxOutputTokens": 64000, @@ -7602,7 +7298,6 @@ "displayName": "Claude Sonnet 4 (latest)", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 216000, "inputLimit": 128000, "maxOutputTokens": 16000, @@ -7622,7 +7317,6 @@ "displayName": "Claude Sonnet 4.5 (latest)", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 168000, "maxOutputTokens": 32000, @@ -7642,7 +7336,6 @@ "displayName": "Claude Sonnet 4.6", "description": "Claude workhorse for coding agents, careful analysis, and production cost control", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 168000, "maxOutputTokens": 32000, @@ -7665,7 +7358,6 @@ "displayName": "Claude Sonnet 5", "description": "Everyday Claude agent model for coding, planning, browsing, and general work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -7687,7 +7379,6 @@ "displayName": "Gemini 3.1 Pro Preview", "description": "Reasoning-first Gemini preview for agentic coding and complex problem solving", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1000000, "inputLimit": 936000, "maxOutputTokens": 64000, @@ -7703,7 +7394,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -7711,7 +7402,6 @@ "displayName": "Gemini 3.5 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 200000, "inputLimit": 128000, "maxOutputTokens": 64000, @@ -7727,7 +7417,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -7735,7 +7425,6 @@ "displayName": "Gemini 3.6 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1000000, "inputLimit": 936000, "maxOutputTokens": 64000, @@ -7759,7 +7448,6 @@ "displayName": "Gemini 3.7 Flash", "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1000000, "inputLimit": 936000, "maxOutputTokens": 64000, @@ -7783,7 +7471,6 @@ "displayName": "GPT-4.1", "description": "Long-lived GPT workhorse for coding, instruction following, and production apps", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 128000, "inputLimit": 128000, "maxOutputTokens": 16384, @@ -7804,7 +7491,6 @@ "displayName": "GPT-5 Mini", "description": "Small GPT-5 for responsive agents, coding help, and everyday automation", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 264000, "inputLimit": 128000, "maxOutputTokens": 64000, @@ -7828,7 +7514,6 @@ "displayName": "GPT-5.2", "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -7849,7 +7534,6 @@ "displayName": "GPT-5.2 Codex", "description": "Code-specialist GPT for repository edits, reviews, and long-running software agents", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -7870,7 +7554,6 @@ "displayName": "GPT-5.3 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -7894,7 +7577,6 @@ "displayName": "GPT-5.4", "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -7918,7 +7600,6 @@ "displayName": "GPT-5.4 mini", "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -7942,7 +7623,6 @@ "displayName": "GPT-5.4 nano", "description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -7963,7 +7643,6 @@ "displayName": "GPT-5.5", "description": "Default frontier GPT for coding, computer use, research, and knowledge work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -7987,7 +7666,6 @@ "displayName": "GPT-5.6 Luna", "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -8011,7 +7689,6 @@ "displayName": "GPT-5.6 Sol", "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -8035,7 +7712,6 @@ "displayName": "GPT-5.6 Terra", "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -8059,7 +7735,6 @@ "displayName": "Grok 4.5", "description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 500000, "inputLimit": 372000, "maxOutputTokens": 128000, @@ -8082,7 +7757,6 @@ "displayName": "Grok 4.6", "description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 500000, "inputLimit": 372000, "maxOutputTokens": 128000, @@ -8106,7 +7780,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 256000, "inputLimit": 224000, "maxOutputTokens": 32000, @@ -8127,7 +7800,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -8149,7 +7821,6 @@ "displayName": "MAI-Code-1-Flash", "description": "Microsoft coding model built for fast, efficient assistance in everyday developer workflows", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 256000, "inputLimit": 128000, "maxOutputTokens": 128000, @@ -8173,7 +7844,6 @@ "displayName": "MAI-Code-1.1-Flash", "description": "Microsoft coding model with native vision support, optimized for fast and efficient software development", "lifecycle": "active", - "docsUrl": "https://docs.github.com/en/copilot", "contextWindow": 256000, "inputLimit": 128000, "maxOutputTokens": 128000, @@ -8198,7 +7868,6 @@ "displayName": "Deep Research Max Preview (Apr-21-2026)", "description": "Maximum-comprehensiveness agentic researcher for multi-step investigation, synthesis, and cited reports", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8209,7 +7878,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text", "image"] } }, @@ -8217,7 +7886,6 @@ "displayName": "Deep Research Preview (Apr-21-2026)", "description": "Agentic model for autonomous multi-step research, synthesis, and cited reports", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8228,7 +7896,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text", "image"] } }, @@ -8236,7 +7904,6 @@ "displayName": "Gemini 2.5 Computer Use Preview 10-2025", "description": "Specialized Gemini 2.5 model for browser-control agents that automate UI tasks", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8255,7 +7922,6 @@ "displayName": "Gemini 2.5 Flash", "description": "Fast Gemini workhorse for multimodal apps where latency and price matter", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8270,7 +7936,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -8278,7 +7944,6 @@ "displayName": "Nano Banana", "description": "Nano Banana image model for fast generation, edits, and character-consistent assets", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 32768, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-06", @@ -8297,7 +7962,6 @@ "displayName": "Gemini 2.5 Flash-Lite", "description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8312,7 +7976,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -8320,7 +7984,6 @@ "displayName": "Gemini 2.5 Flash Preview TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 8192, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-01", @@ -8339,7 +8002,6 @@ "displayName": "Gemini 2.5 Pro", "description": "Google's proven reasoning model for coding, math, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8351,7 +8013,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -8359,7 +8021,6 @@ "displayName": "Gemini 2.5 Pro Preview TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 8192, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-01", @@ -8378,7 +8039,6 @@ "displayName": "Gemini 3 Flash Preview", "description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8393,7 +8053,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8401,7 +8061,6 @@ "displayName": "Nano Banana Pro", "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -8423,7 +8082,6 @@ "displayName": "Nano Banana Pro", "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -8442,7 +8100,6 @@ "displayName": "Nano Banana 2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 65536, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8456,7 +8113,7 @@ "efforts": ["minimal", "high"] }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text", "image"] } }, @@ -8464,7 +8121,6 @@ "displayName": "Nano Banana 2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 65536, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8486,7 +8142,6 @@ "displayName": "Gemini 3.1 Flash Lite", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8501,7 +8156,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8509,7 +8164,6 @@ "displayName": "Nano Banana 2 Lite", "description": "Fastest, most cost-efficient Gemini image model for high-volume 1K generation and editing", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 65536, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8532,7 +8186,6 @@ "displayName": "Gemini 3.1 Flash Lite Preview", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8547,7 +8200,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8555,7 +8208,6 @@ "displayName": "Gemini 3.1 Flash Live Preview", "description": "High-quality, low-latency Live API model for real-time dialogue and voice-first AI applications", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8570,7 +8222,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text", "audio"] } }, @@ -8578,7 +8230,6 @@ "displayName": "Gemini 3.1 Flash TTS Preview", "description": "Low-latency speech generation with steerable prompts and expressive audio tags", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 8192, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-01", @@ -8597,7 +8248,6 @@ "displayName": "Gemini 3.1 Pro Preview", "description": "Reasoning-first Gemini preview for agentic coding and complex problem solving", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8612,7 +8262,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8620,7 +8270,6 @@ "displayName": "Gemini 3.1 Pro Preview Custom Tools", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8635,7 +8284,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8643,7 +8292,6 @@ "displayName": "Gemini 3.5 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -8658,7 +8306,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8666,7 +8314,6 @@ "displayName": "Gemini 3.5 Flash Lite", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -8681,7 +8328,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8689,7 +8336,6 @@ "displayName": "Gemini 3.5 Live Translate Preview", "description": "Low-latency audio-to-audio model for real-time speech translation across 70+ languages", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 16384, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -8708,7 +8354,6 @@ "displayName": "Gemini 3.6 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -8723,7 +8368,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8731,7 +8376,6 @@ "displayName": "Gemini 3.7 Flash", "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -8746,7 +8390,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8754,7 +8398,6 @@ "displayName": "Gemini Embedding 001", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 2048, "maxOutputTokens": 1, "knowledgeCutoff": "2025-05", @@ -8773,7 +8416,6 @@ "displayName": "Gemini Embedding 2", "description": "Multimodal embedding model mapping text, images, video, audio, and PDFs into a unified embedding space", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 8192, "maxOutputTokens": 1, "knowledgeCutoff": "2025-11", @@ -8784,7 +8426,7 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -8792,7 +8434,6 @@ "displayName": "Gemini Flash Latest", "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -8807,7 +8448,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8815,7 +8456,6 @@ "displayName": "Gemini Flash-Lite Latest", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -8830,7 +8470,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -8838,7 +8478,6 @@ "displayName": "Gemini Omni Flash Preview", "description": "Video generation and editing model for fast, conversational text- and image-to-video workflows", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "lastUpdated": "2026-06-30", @@ -8848,38 +8487,14 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image"], - "output": [] - } - }, - "gemini-robotics-er-1.6-preview": { - "displayName": "Gemini Robotics-ER 1.6 Preview", - "description": "Vision-language model for embodied reasoning: spatial understanding, task planning, and physical-world agentic robotics", - "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", - "contextWindow": 131072, - "maxOutputTokens": 65536, - "knowledgeCutoff": "2025-01", - "structuredOutput": true, - "lastUpdated": "2026-04-14", - "capabilities": { - "vision": true, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "toggle": true - }, - "modalities": { - "input": ["text", "image", "audio"], - "output": ["text"] + "input": ["text", "image", "video"], + "output": ["video"] } }, "gemma-4-26b-a4b-it": { "displayName": "Gemma 4 26B A4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -8901,7 +8516,6 @@ "displayName": "Gemma 4 31B IT", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -8923,7 +8537,6 @@ "displayName": "Lyria 3 Clip Preview", "description": "Music generation model for short 30-second clips, loops, and previews from text or image prompts", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "structuredOutput": false, @@ -8943,7 +8556,6 @@ "displayName": "Lyria 3 Pro Preview", "description": "Music generation model for full-length songs from text or images with vocals and structure", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "structuredOutput": false, @@ -8963,7 +8575,6 @@ "displayName": "Veo 3.1 fast", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 480, "maxOutputTokens": 8192, "lastUpdated": "2026-01-01", @@ -8973,15 +8584,14 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image"], - "output": [] + "input": ["text", "image", "video"], + "output": ["video"] } }, "veo-3.1-generate-preview": { "displayName": "Veo 3.1", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "beta", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 480, "maxOutputTokens": 8192, "lastUpdated": "2026-01", @@ -8992,14 +8602,13 @@ }, "modalities": { "input": ["text", "image"], - "output": [] + "output": ["video"] } }, "veo-3.1-lite-generate-preview": { "displayName": "Veo 3.1 lite", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://ai.google.dev/gemini-api/docs/models", "contextWindow": 480, "maxOutputTokens": 8192, "lastUpdated": "2026-03-31", @@ -9010,7 +8619,7 @@ }, "modalities": { "input": ["text", "image"], - "output": [] + "output": ["video"] } } }, @@ -9019,7 +8628,6 @@ "displayName": "ALLaM-2-7b", "description": "ALLaM-2-7b instruction tuned model by SDAIA", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 4096, "maxOutputTokens": 4096, "lastUpdated": "2025-01-23", @@ -9038,7 +8646,6 @@ "displayName": "Canopy Labs Orpheus Arabic Saudi", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "beta", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 4000, "maxOutputTokens": 50000, "lastUpdated": "2025-12-16", @@ -9056,7 +8663,6 @@ "displayName": "Canopy Labs Orpheus V1 English", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "beta", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 4000, "maxOutputTokens": 50000, "lastUpdated": "2025-12-19", @@ -9074,7 +8680,6 @@ "displayName": "Compound", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 8192, "lastUpdated": "2025-09-04", @@ -9092,7 +8697,6 @@ "displayName": "Compound Mini", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 8192, "lastUpdated": "2025-09-04", @@ -9110,7 +8714,6 @@ "displayName": "Llama 3.1 8B", "description": "Compact Llama instruction model for fast chat and local deployment", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2023-12", @@ -9129,7 +8732,6 @@ "displayName": "Llama 3.3 70B", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2023-12", @@ -9148,7 +8750,6 @@ "displayName": "Llama Prompt Guard 2 22M", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "beta", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 512, "maxOutputTokens": 512, "lastUpdated": "2025-05-29", @@ -9166,7 +8767,6 @@ "displayName": "Prompt Guard 2 86M", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "beta", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 512, "maxOutputTokens": 512, "lastUpdated": "2025-05-29", @@ -9184,7 +8784,6 @@ "displayName": "GPT OSS 120B", "description": "Open GPT reasoning model for self-hosted agents and controllable deployments", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "structuredOutput": true, @@ -9206,7 +8805,6 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "structuredOutput": true, @@ -9228,7 +8826,6 @@ "displayName": "Safety GPT OSS 20B", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "beta", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 65536, "structuredOutput": true, @@ -9250,7 +8847,6 @@ "displayName": "Qwen3.6 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 131072, "maxOutputTokens": 16384, "structuredOutput": true, @@ -9268,11 +8864,31 @@ "output": ["text"] } }, + "qwen/qwen3.8-27b": { + "displayName": "Qwen3.8 27B", + "description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding", + "lifecycle": "active", + "contextWindow": 131042, + "maxOutputTokens": 16384, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "default", "low", "medium", "high"] + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "whisper-large-v3": { "displayName": "Whisper", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-09-05", @@ -9290,7 +8906,6 @@ "displayName": "Whisper Large V3 Turbo", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://console.groq.com/docs/models", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2024-10-01", @@ -9310,7 +8925,6 @@ "displayName": "DeepSeek-R1", "description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 64000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-07", @@ -9330,7 +8944,6 @@ "displayName": "DeepSeek-R1-0528", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 163840, "maxOutputTokens": 163840, "knowledgeCutoff": "2025-05", @@ -9349,7 +8962,6 @@ "displayName": "DeepSeek-V3", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 64000, "maxOutputTokens": 8192, "structuredOutput": true, @@ -9368,7 +8980,6 @@ "displayName": "DeepSeek V3 0324", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 163840, "maxOutputTokens": 163840, "structuredOutput": true, @@ -9387,7 +8998,6 @@ "displayName": "DeepSeek-V3.1", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 8192, "structuredOutput": true, @@ -9406,7 +9016,6 @@ "displayName": "DeepSeek-V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 163840, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-07", @@ -9425,7 +9034,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 1048576, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -9445,7 +9053,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 1048576, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -9468,7 +9075,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 1048576, "maxOutputTokens": 393216, "knowledgeCutoff": "2025-05", @@ -9491,11 +9097,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "Flagship DeepSeek model for coding, reasoning, and agentic work", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 1000000, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -9513,7 +9118,6 @@ "displayName": "Gemma 4 26B A4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -9532,7 +9136,6 @@ "displayName": "Gemma 4 31B IT", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -9551,7 +9154,6 @@ "displayName": "Llama-3.1-8B-Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -9571,7 +9173,6 @@ "displayName": "Llama-3.3-70B-Instruct", "description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -9591,9 +9192,8 @@ "displayName": "MiniMax-M2", "description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 204800, - "maxOutputTokens": 128000, + "maxOutputTokens": 131072, "lastUpdated": "2025-10-27", "capabilities": { "vision": false, @@ -9609,7 +9209,6 @@ "displayName": "MiniMax-M2.1", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-10", @@ -9628,7 +9227,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-12", @@ -9646,7 +9244,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 204800, "maxOutputTokens": 131072, "structuredOutput": true, @@ -9665,9 +9262,8 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 524288, - "maxOutputTokens": 128000, + "maxOutputTokens": 512000, "structuredOutput": true, "lastUpdated": "2026-06-01", "capabilities": { @@ -9684,7 +9280,6 @@ "displayName": "Kimi-K2-Instruct", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-10", @@ -9703,7 +9298,6 @@ "displayName": "Kimi-K2-Instruct-0905", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-10", @@ -9722,7 +9316,6 @@ "displayName": "Kimi-K2-Thinking", "description": "Kimi reasoning model for long-horizon research, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-08", @@ -9741,7 +9334,6 @@ "displayName": "Kimi-K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -9752,7 +9344,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -9760,7 +9352,6 @@ "displayName": "Kimi-K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -9771,7 +9362,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -9779,7 +9370,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -9799,7 +9389,6 @@ "displayName": "Kimi K3", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -9821,7 +9410,6 @@ "displayName": "GPT OSS 120B", "description": "Open GPT reasoning model for self-hosted agents and controllable deployments", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": true, @@ -9843,7 +9431,6 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": true, @@ -9865,7 +9452,6 @@ "displayName": "Qwen2.5-Coder-32B-Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 8192, "structuredOutput": true, @@ -9884,7 +9470,6 @@ "displayName": "Qwen3 235B-A22B", "description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 40960, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -9904,7 +9489,6 @@ "displayName": "Qwen3 235B-A22B Instruct 2507", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 16384, "structuredOutput": true, @@ -9923,7 +9507,6 @@ "displayName": "Qwen3-235B-A22B-Thinking-2507", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -9942,7 +9525,6 @@ "displayName": "Qwen3 30B A3B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 40960, "maxOutputTokens": 16384, "structuredOutput": true, @@ -9961,7 +9543,6 @@ "displayName": "Qwen3 32B", "description": "Dense open Qwen model for self-hosted chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -9981,7 +9562,6 @@ "displayName": "Qwen3-Coder 30B-A3B Instruct", "description": "Smaller Qwen coder for efficient local agents and repo-level fixes", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -10001,7 +9581,6 @@ "displayName": "Qwen3-Coder-480B-A35B-Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 66536, "knowledgeCutoff": "2025-04", @@ -10020,7 +9599,6 @@ "displayName": "Qwen3-Coder-Next", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -10039,7 +9617,6 @@ "displayName": "Qwen 3 Embedding 4B", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 32000, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-12", @@ -10058,7 +9635,6 @@ "displayName": "Qwen 3 Embedding 8B", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 32000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-12", @@ -10077,7 +9653,6 @@ "displayName": "Qwen3-Next-80B-A3B-Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 66536, "knowledgeCutoff": "2025-04", @@ -10096,7 +9671,6 @@ "displayName": "Qwen3-Next-80B-A3B-Thinking", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -10115,7 +9689,6 @@ "displayName": "Qwen3 VL 235B A22B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -10135,7 +9708,6 @@ "displayName": "Qwen3 VL 235B A22B Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -10158,7 +9730,6 @@ "displayName": "Qwen3.5 122B-A10B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -10177,7 +9748,6 @@ "displayName": "Qwen3.5 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -10196,7 +9766,6 @@ "displayName": "Qwen3.5 35B-A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -10215,7 +9784,6 @@ "displayName": "Qwen3.5-397B-A17B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -10237,7 +9805,6 @@ "displayName": "Qwen3.5 9B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -10256,7 +9823,6 @@ "displayName": "Qwen3.6 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -10275,7 +9841,6 @@ "displayName": "Qwen3.6 35B-A3B", "description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -10294,7 +9859,6 @@ "displayName": "Qwen3.8 2.4T A95B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -10312,11 +9876,31 @@ "output": ["text"] } }, + "Qwen/Qwen3.8-27B": { + "displayName": "Qwen3.8 27B", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 262144, + "maxOutputTokens": 32768, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"] + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "stepfun-ai/Step-3.5-Flash": { "displayName": "Step 3.5 Flash", "description": "StepFun flash lane for quick multimodal reasoning and coding assistance", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 256000, "knowledgeCutoff": "2025-01", @@ -10335,7 +9919,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 256000, "knowledgeCutoff": "2026-03-01", @@ -10349,7 +9932,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -10357,7 +9940,6 @@ "displayName": "Hy3", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 64000, "structuredOutput": true, @@ -10379,7 +9961,6 @@ "displayName": "Inkling", "description": "Multimodal model for analyzing text, images, documents, and rich media", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 1048576, "maxOutputTokens": 1048576, "structuredOutput": true, @@ -10401,7 +9982,6 @@ "displayName": "Inkling Small", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 524288, "maxOutputTokens": 1048576, "lastUpdated": "2026-07-30", @@ -10419,7 +9999,6 @@ "displayName": "MiMo-V2-Flash", "description": "MiMo flash model for fast multimodal assistance and agent workflows", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-12", @@ -10438,7 +10017,6 @@ "displayName": "MiMo-V2.5", "description": "MiMo model for long-context reasoning, perception, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -10462,7 +10040,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -10486,7 +10063,6 @@ "displayName": "GLM-4.5", "description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-04", @@ -10505,7 +10081,6 @@ "displayName": "GLM-4.5-Air", "description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-04", @@ -10524,7 +10099,6 @@ "displayName": "GLM-4.5V", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 65536, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -10543,7 +10117,6 @@ "displayName": "GLM-4.6", "description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -10562,7 +10135,6 @@ "displayName": "GLM-4.6V-Flash", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 131072, "maxOutputTokens": 32768, "lastUpdated": "2025-12-08", @@ -10583,7 +10155,6 @@ "displayName": "GLM-4.7", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -10602,7 +10173,6 @@ "displayName": "GLM-4.7-Flash", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 200000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-04", @@ -10622,7 +10192,6 @@ "displayName": "GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 202752, "maxOutputTokens": 131072, "lastUpdated": "2026-02-11", @@ -10640,7 +10209,6 @@ "displayName": "GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 202752, "maxOutputTokens": 131072, "lastUpdated": "2026-04-03", @@ -10658,7 +10226,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://huggingface.co/docs/inference-providers", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -10672,6 +10239,48 @@ "input": ["text"], "output": ["text"] } + }, + "zai-org/GLM-5.3": { + "displayName": "GLM-5.3", + "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "zai-org/GLM-5.3-Flash": { + "displayName": "GLM-5.3-Flash", + "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } } }, "kimi-coding-plan": { @@ -10679,7 +10288,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://www.kimi.com/code/docs/en/third-party-tools/other-coding-agents.html", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -10695,7 +10303,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -10703,7 +10311,6 @@ "displayName": "Kimi K3-256K", "description": "256K-context version of Kimi K3, reducing token consumption for shorter coding sessions", "lifecycle": "active", - "docsUrl": "https://www.kimi.com/code/docs/en/third-party-tools/other-coding-agents.html", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -10726,7 +10333,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://www.kimi.com/code/docs/en/third-party-tools/other-coding-agents.html", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -10739,7 +10345,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -10747,7 +10353,6 @@ "displayName": "Kimi For Coding HighSpeed", "description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops", "lifecycle": "active", - "docsUrl": "https://www.kimi.com/code/docs/en/third-party-tools/other-coding-agents.html", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -10760,7 +10365,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -10770,9 +10375,8 @@ "displayName": "MiniMax-M2", "description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/guides/quickstart", - "contextWindow": 196608, - "maxOutputTokens": 128000, + "contextWindow": 204800, + "maxOutputTokens": 131072, "lastUpdated": "2025-10-27", "capabilities": { "vision": false, @@ -10788,7 +10392,6 @@ "displayName": "MiniMax-M2.1", "description": "Earlier MiniMax agent model for practical coding and productivity tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2025-12-23", @@ -10806,7 +10409,6 @@ "displayName": "MiniMax-M2.5", "description": "Prior MiniMax coding model for agent workflows, office edits, and automation", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-12", @@ -10824,7 +10426,6 @@ "displayName": "MiniMax-M2.5-highspeed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-13", @@ -10842,7 +10443,6 @@ "displayName": "MiniMax-M2.7", "description": "Open MiniMax flagship for coding agents, office automation, and complex environments", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -10860,7 +10460,6 @@ "displayName": "MiniMax-M2.7-highspeed", "description": "Low-latency M2.7 variant for interactive coding plans and agent loops", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -10878,9 +10477,8 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/guides/quickstart", - "contextWindow": 1000000, - "maxOutputTokens": 128000, + "contextWindow": 1048576, + "maxOutputTokens": 512000, "lastUpdated": "2026-06-25", "capabilities": { "vision": true, @@ -10891,7 +10489,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -10901,9 +10499,8 @@ "displayName": "MiniMax-M2", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimaxi.com/docs/guides/quickstart", - "contextWindow": 196608, - "maxOutputTokens": 128000, + "contextWindow": 204800, + "maxOutputTokens": 131072, "lastUpdated": "2025-10-27", "capabilities": { "vision": false, @@ -10919,7 +10516,6 @@ "displayName": "MiniMax-M2.1", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimaxi.com/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2025-12-23", @@ -10937,7 +10533,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimaxi.com/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-12", @@ -10955,7 +10550,6 @@ "displayName": "MiniMax-M2.5-highspeed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://platform.minimaxi.com/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-13", @@ -10973,7 +10567,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimaxi.com/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -10991,7 +10584,6 @@ "displayName": "MiniMax-M2.7-highspeed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://platform.minimaxi.com/docs/guides/quickstart", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -11009,9 +10601,8 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimaxi.com/docs/guides/quickstart", - "contextWindow": 1000000, - "maxOutputTokens": 128000, + "contextWindow": 1048576, + "maxOutputTokens": 512000, "lastUpdated": "2026-06-25", "capabilities": { "vision": true, @@ -11022,7 +10613,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -11032,9 +10623,8 @@ "displayName": "MiniMax-M2", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/token-plan/intro", - "contextWindow": 196608, - "maxOutputTokens": 128000, + "contextWindow": 204800, + "maxOutputTokens": 131072, "lastUpdated": "2025-10-27", "isFree": true, "capabilities": { @@ -11051,7 +10641,6 @@ "displayName": "MiniMax-M2.1", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/token-plan/intro", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2025-12-23", @@ -11070,7 +10659,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/token-plan/intro", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-12", @@ -11089,7 +10677,6 @@ "displayName": "MiniMax-M2.5-highspeed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/token-plan/intro", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-13", @@ -11108,7 +10695,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/token-plan/intro", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -11127,7 +10713,6 @@ "displayName": "MiniMax-M2.7-highspeed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/token-plan/intro", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-03-18", @@ -11146,9 +10731,8 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://platform.minimax.io/docs/token-plan/intro", - "contextWindow": 1000000, - "maxOutputTokens": 128000, + "contextWindow": 1048576, + "maxOutputTokens": 512000, "lastUpdated": "2026-06-25", "isFree": true, "capabilities": { @@ -11160,7 +10744,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -11170,7 +10754,6 @@ "displayName": "Codestral (latest)", "description": "Mistral code model for completions, refactors, and developer IDE workflows", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 256000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-10", @@ -11189,7 +10772,6 @@ "displayName": "Devstral 2", "description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-12", @@ -11208,7 +10790,6 @@ "displayName": "Devstral 2", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-12", @@ -11227,7 +10808,6 @@ "displayName": "Devstral Medium", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05", @@ -11246,7 +10826,6 @@ "displayName": "Devstral 2 (latest)", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-12", @@ -11265,7 +10844,6 @@ "displayName": "Devstral Small 2505", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05", @@ -11284,7 +10862,6 @@ "displayName": "Devstral Small", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05", @@ -11303,7 +10880,6 @@ "displayName": "Devstral Small 2", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2025-12", @@ -11323,7 +10899,6 @@ "displayName": "Magistral Medium (latest)", "description": "Mistral reasoning model for transparent analysis, math, and complex decisions", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-06", @@ -11342,7 +10917,6 @@ "displayName": "Magistral Small", "description": "Mistral reasoning model for transparent analysis, math, and complex decisions", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-06", @@ -11361,7 +10935,6 @@ "displayName": "Ministral 3B (latest)", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-10", @@ -11380,7 +10953,6 @@ "displayName": "Ministral 8B (latest)", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-10", @@ -11399,7 +10971,6 @@ "displayName": "Mistral Embed", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 8000, "maxOutputTokens": 3072, "lastUpdated": "2023-12-11", @@ -11417,7 +10988,6 @@ "displayName": "Mistral Large 2.1", "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-11", @@ -11436,7 +11006,6 @@ "displayName": "Mistral Large 3", "description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-11", @@ -11455,7 +11024,6 @@ "displayName": "Mistral Large (latest)", "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-11", @@ -11474,7 +11042,6 @@ "displayName": "Mistral Medium 3", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-05", @@ -11493,7 +11060,6 @@ "displayName": "Mistral Medium 3.1", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-05", @@ -11512,7 +11078,6 @@ "displayName": "Mistral Medium 3.5", "description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -11534,7 +11099,6 @@ "displayName": "Mistral Medium (latest)", "description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -11556,7 +11120,6 @@ "displayName": "Mistral Nemo", "description": "Efficient Mistral-NVIDIA open model for multilingual chat and local deployment", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-07", @@ -11575,7 +11138,6 @@ "displayName": "Mistral Small 3.2", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-03", @@ -11594,7 +11156,6 @@ "displayName": "Mistral Small 4", "description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2025-06", @@ -11616,7 +11177,6 @@ "displayName": "Mistral Small (latest)", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2025-06", @@ -11638,7 +11198,6 @@ "displayName": "Mistral 7B", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 8000, "maxOutputTokens": 8000, "knowledgeCutoff": "2023-12", @@ -11657,7 +11216,6 @@ "displayName": "Open Mistral Nemo", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-07", @@ -11676,7 +11234,6 @@ "displayName": "Mixtral 8x22B", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 64000, "maxOutputTokens": 64000, "knowledgeCutoff": "2024-04", @@ -11695,7 +11252,6 @@ "displayName": "Mixtral 8x7B", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 32000, "maxOutputTokens": 32000, "knowledgeCutoff": "2024-01", @@ -11714,7 +11270,6 @@ "displayName": "Pixtral 12B", "description": "Mistral vision-language model for image understanding and multimodal chat", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-09", @@ -11733,7 +11288,6 @@ "displayName": "Pixtral Large (latest)", "description": "Mistral's larger vision model for document-heavy image understanding and chat", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-11", @@ -11752,7 +11306,6 @@ "displayName": "Voxtral Mini (latest)", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-02-01", @@ -11770,7 +11323,6 @@ "displayName": "Voxtral Mini TTS (latest)", "description": "Multilingual text-to-speech model with zero-shot voice cloning", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-01", @@ -11788,7 +11340,6 @@ "displayName": "Voxtral Small (latest)", "description": "Instruct model with native audio input for speech understanding and tool use", "lifecycle": "active", - "docsUrl": "https://docs.mistral.ai/getting-started/models/", "contextWindow": 32000, "maxOutputTokens": 32000, "lastUpdated": "2025-07-15", @@ -11801,6 +11352,27 @@ "input": ["text", "audio"], "output": ["text"] } + }, + "zai-glm-5-2": { + "displayName": "GLM-5.2", + "description": "Open flagship GLM for long-horizon coding agents and million-token context work", + "lifecycle": "beta", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-06-13", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } } }, "moonshot": { @@ -11808,7 +11380,6 @@ "displayName": "Kimi K2 0711", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-10", @@ -11827,7 +11398,6 @@ "displayName": "Kimi K2 0905", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-10", @@ -11846,7 +11416,6 @@ "displayName": "Kimi K2 Thinking", "description": "Thinking Kimi model for slower research passes, planning, and hard technical questions", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-08", @@ -11865,7 +11434,6 @@ "displayName": "Kimi K2 Thinking Turbo", "description": "Kimi reasoning model for long-horizon research, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-08", @@ -11884,7 +11452,6 @@ "displayName": "Kimi K2 Turbo", "description": "Fast Kimi model for responsive chat, coding help, and agent loops", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-10", @@ -11903,7 +11470,6 @@ "displayName": "Kimi K2.5", "description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -11918,7 +11484,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -11926,7 +11492,6 @@ "displayName": "Kimi K2.6", "description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -11941,7 +11506,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -11949,7 +11514,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -11961,7 +11525,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -11969,7 +11533,6 @@ "displayName": "Kimi K2.7 Code HighSpeed", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -11981,7 +11544,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -11989,7 +11552,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://platform.moonshot.cn/docs/api/chat", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -12004,7 +11566,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -12014,7 +11576,6 @@ "displayName": "dracarys-llama-3.1-70b-instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-05-22", @@ -12033,7 +11594,6 @@ "displayName": "BGE M3", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 8192, "maxOutputTokens": 1024, "lastUpdated": "2026-04-30", @@ -12052,7 +11612,6 @@ "displayName": "FLUX.1-Kontext-dev", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 40960, "maxOutputTokens": 40960, "lastUpdated": "2025-08-12", @@ -12071,7 +11630,6 @@ "displayName": "FLUX.1-schnell", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 77, "inputLimit": 77, "maxOutputTokens": 0, @@ -12093,7 +11651,6 @@ "displayName": "FLUX.2 Klein 4B", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 40960, "maxOutputTokens": 40960, "knowledgeCutoff": "2025-06", @@ -12113,7 +11670,6 @@ "displayName": "FLUX.1-dev", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 4096, "maxOutputTokens": 0, "knowledgeCutoff": "2024-08", @@ -12133,7 +11689,6 @@ "displayName": "ByteDance-Seed/Seed-OSS-36B-Instruct", "description": "Tool-capable chat model for instruction following and agentic application workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -12153,7 +11708,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 1048576, "maxOutputTokens": 393216, "knowledgeCutoff": "2025-05", @@ -12172,11 +11726,33 @@ "output": ["text"] } }, + "deepseek-ai/deepseek-v4-flash-0731": { + "displayName": "DeepSeek V4 Flash 0731", + "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 384000, + "knowledgeCutoff": "2025-05", + "structuredOutput": true, + "lastUpdated": "2026-07-31", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "deepseek-ai/deepseek-v4-pro": { "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 1048576, "maxOutputTokens": 393216, "knowledgeCutoff": "2025-05", @@ -12195,11 +11771,32 @@ "output": ["text"] } }, + "deepseek-ai/deepseek-v4-pro-0813": { + "displayName": "DeepSeek V4 Pro 0813", + "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 384000, + "structuredOutput": true, + "lastUpdated": "2026-08-22", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "google/gemma-2-2b-it": { "displayName": "Gemma 2 2b It", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "structuredOutput": true, @@ -12219,7 +11816,6 @@ "displayName": "Gemma 3 12B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2025-03-12", @@ -12238,7 +11834,6 @@ "displayName": "Gemma 3 4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2025-03-12", @@ -12257,7 +11852,6 @@ "displayName": "Gemma 3n E2b It", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-06", @@ -12278,7 +11872,6 @@ "displayName": "Gemma 3n E4b It", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-06", @@ -12299,7 +11892,6 @@ "displayName": "Gemma-4-31B-IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 256000, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-01", @@ -12314,7 +11906,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -12322,7 +11914,6 @@ "displayName": "paligemma", "description": "Gemini multimodal model for text, image, audio, video, and document tasks", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2024-08-26", @@ -12341,7 +11932,6 @@ "displayName": "esm2-650m", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-03-10", @@ -12360,7 +11950,6 @@ "displayName": "esmfold", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-06-12", @@ -12379,7 +11968,6 @@ "displayName": "Llama 3.1 70b Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "structuredOutput": true, @@ -12399,7 +11987,6 @@ "displayName": "Llama 3.1 8B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 16000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -12419,7 +12006,6 @@ "displayName": "Llama 3.2 11b Vision Instruct", "description": "Open Llama multimodal model for image understanding and text reasoning", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -12440,7 +12026,6 @@ "displayName": "Llama 3.2 1b Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -12461,7 +12046,6 @@ "displayName": "Llama 3.2 3B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "maxOutputTokens": 32000, "structuredOutput": true, @@ -12481,7 +12065,6 @@ "displayName": "Llama-3.2-90B-Vision-Instruct", "description": "Open Llama multimodal model for image understanding and text reasoning", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "knowledgeCutoff": "2023-12", @@ -12501,7 +12084,6 @@ "displayName": "Llama 3.3 70b Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "structuredOutput": true, @@ -12521,7 +12103,6 @@ "displayName": "Llama 4 Maverick 17b 128e Instruct", "description": "Open multimodal Llama model for strong reasoning and fast responses", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-02", @@ -12542,7 +12123,6 @@ "displayName": "Llama Guard 4 12B", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 16384, "lastUpdated": "2026-04-30", @@ -12561,7 +12141,6 @@ "displayName": "Muse Glimmer 30B", "description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2026-01-04", @@ -12585,7 +12164,6 @@ "displayName": "Phi-4-Mini", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-12", @@ -12605,7 +12183,6 @@ "displayName": "Phi 4 Multimodal", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "inputLimit": 128000, "maxOutputTokens": 16384, @@ -12626,7 +12203,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-04-11", @@ -12645,7 +12221,6 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 1000000, "maxOutputTokens": 16384, "lastUpdated": "2026-06-01", @@ -12659,7 +12234,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -12667,7 +12242,6 @@ "displayName": "Magistral Small 2506", "description": "Mistral reasoning model for transparent analysis, math, and complex decisions", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "inputLimit": 32768, "maxOutputTokens": 32768, @@ -12688,7 +12262,6 @@ "displayName": "Ministral 3 14B Instruct 2512", "description": "Compact Mistral VLM for chat and instruction-based workloads", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 16384, "lastUpdated": "2025-12-02", @@ -12707,7 +12280,6 @@ "displayName": "Mistral-7B-Instruct-v0.3", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 65536, "maxOutputTokens": 65536, "structuredOutput": true, @@ -12727,7 +12299,6 @@ "displayName": "Mistral Large 3 675B Instruct 2512", "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -12748,7 +12319,6 @@ "displayName": "Mistral Medium 3", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "inputLimit": 131072, "maxOutputTokens": 32768, @@ -12769,7 +12339,6 @@ "displayName": "Mistral Medium 3.5", "description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -12792,7 +12361,6 @@ "displayName": "mistral-nemotron", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-06-12", @@ -12811,7 +12379,6 @@ "displayName": "mistral-small-4-119b-2603", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "structuredOutput": true, @@ -12834,7 +12401,6 @@ "displayName": "Mistral: Mixtral 8x22B Instruct", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 65536, "maxOutputTokens": 13108, "lastUpdated": "2024-04-17", @@ -12853,7 +12419,6 @@ "displayName": "Mistral: Mixtral 8x7B Instruct", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2026-03-15", @@ -12872,7 +12437,6 @@ "displayName": "Kimi K2 0905", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "deprecated", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-10", @@ -12892,7 +12456,6 @@ "displayName": "Kimi K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "deprecated", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -12908,7 +12471,30 @@ "efforts": ["none", "minimal", "low", "medium", "high", "xhigh", "max"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "moonshotai/kimi-k3": { + "displayName": "Kimi K3", + "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-07-16", + "isFree": true, + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"], + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -12916,7 +12502,6 @@ "displayName": "Active Speaker Detection", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "lastUpdated": "2026-04-16", @@ -12927,7 +12512,7 @@ "functionCalling": false }, "modalities": { - "input": [], + "input": ["video"], "output": ["text"] } }, @@ -12935,7 +12520,6 @@ "displayName": "bevformer", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-07-20", @@ -12946,7 +12530,7 @@ "functionCalling": false }, "modalities": { - "input": [], + "input": ["video"], "output": ["text"] } }, @@ -12954,7 +12538,6 @@ "displayName": "cosmos-predict1-5b", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "lastUpdated": "2025-03-18", @@ -12965,15 +12548,14 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image"], - "output": [] + "input": ["text", "image", "video"], + "output": ["video"] } }, "nvidia/cosmos-reason2-8b": { "displayName": "Cosmos Reason2 8B", "description": "Vision language model for physical-world understanding with structured reasoning on video and images", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2025-12-01", @@ -12984,7 +12566,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -12992,7 +12574,6 @@ "displayName": "cosmos-transfer1-7b", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "lastUpdated": "2025-06-30", @@ -13003,15 +12584,14 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image"], - "output": [] + "input": ["text", "image", "video"], + "output": ["video"] } }, "nvidia/cosmos-transfer2_5-2b": { "displayName": "cosmos-transfer2.5-2b", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "lastUpdated": "2026-02-26", @@ -13022,15 +12602,14 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image"], - "output": [] + "input": ["text", "image", "video"], + "output": ["video"] } }, "nvidia/gliner-pii": { "displayName": "gliner-pii", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2026-03-03", @@ -13049,7 +12628,6 @@ "displayName": "llama-3_2-nemoretriever-300m-embed-v1", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "maxOutputTokens": 2048, "lastUpdated": "2025-07-24", @@ -13068,7 +12646,6 @@ "displayName": "Llama 3.1 Nemotron 70B Instruct", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-04-15", @@ -13087,7 +12664,6 @@ "displayName": "Llama 3.1 Nemotron Nano 8B v1", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2025-03-18", @@ -13109,7 +12685,6 @@ "displayName": "Llama 3.1 Nemotron Nano VL 8B v1", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "maxOutputTokens": 16384, "lastUpdated": "2025-04-10", @@ -13128,7 +12703,6 @@ "displayName": "llama-3.1-nemotron-safety-guard-8b-v3", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2025-10-28", @@ -13147,7 +12721,6 @@ "displayName": "Llama 3.1 Nemotron Ultra 253B", "description": "Flagship Nemotron model for high-throughput reasoning and complex agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 16384, "lastUpdated": "2025-04-07", @@ -13169,7 +12742,6 @@ "displayName": "Llama 3.3 Nemotron Super 49B v1", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 65536, "lastUpdated": "2025-04-07", @@ -13191,7 +12763,6 @@ "displayName": "Llama 3.3 Nemotron Super 49B v1.5", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 65536, "lastUpdated": "2025-07-25", @@ -13213,7 +12784,6 @@ "displayName": "llama-nemotron-embed-vl-1b-v2", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "maxOutputTokens": 2048, "lastUpdated": "2026-02-10", @@ -13232,7 +12802,6 @@ "displayName": "llama-nemotron-rerank-vl-1b-v2", "description": "Reranking model for improving retrieval quality in search and recommendation systems", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2026-03-31", @@ -13251,7 +12820,6 @@ "displayName": "magpie-tts-zeroshot", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "lastUpdated": "2025-06-12", @@ -13270,7 +12838,6 @@ "displayName": "nemotron-3-content-safety", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2026-04-16", @@ -13289,7 +12856,6 @@ "displayName": "nemotron-3-nano-30b-a3b", "description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-09", @@ -13312,7 +12878,6 @@ "displayName": "Nemotron 3 Nano Omni", "description": "Open Nemotron omni model combining reasoning with text, vision, and audio", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 256000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -13327,7 +12892,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -13335,7 +12900,6 @@ "displayName": "Nemotron 3 Super", "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-04", @@ -13357,7 +12921,6 @@ "displayName": "Nemotron 3 Ultra 550B A55B", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -13379,7 +12942,6 @@ "displayName": "Nemotron 3.5 Lightning 30B A3B", "description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -13402,7 +12964,6 @@ "displayName": "nemotron-content-safety-reasoning-4b", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2026-01-22", @@ -13421,7 +12982,6 @@ "displayName": "nemotron-mini-4b-instruct", "description": "Compact Nemotron model for efficient reasoning and deployable AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2024-08-26", @@ -13440,7 +13000,6 @@ "displayName": "Nemotron Nano 12B v2 VL", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 128000, "lastUpdated": "2025-10-28", @@ -13451,7 +13010,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -13459,7 +13018,6 @@ "displayName": "nemotron-voicechat", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2026-03-16", @@ -13478,7 +13036,6 @@ "displayName": "nv-embed-v1", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "maxOutputTokens": 2048, "lastUpdated": "2025-07-22", @@ -13497,7 +13054,6 @@ "displayName": "nv-embedcode-7b-v1", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 32768, "maxOutputTokens": 2048, "lastUpdated": "2025-05-29", @@ -13516,7 +13072,6 @@ "displayName": "nvidia-nemotron-nano-9b-v2", "description": "Compact Nemotron model for efficient reasoning and deployable AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-09", @@ -13539,7 +13094,6 @@ "displayName": "rerank-qa-mistral-4b", "description": "Reranking model for improving retrieval quality in search and recommendation systems", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2025-01-17", @@ -13558,7 +13112,6 @@ "displayName": "riva-translate-4b-instruct-v1_1", "description": "Translation model for multilingual conversion, localization, and cross-language workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2025-12-12", @@ -13577,7 +13130,6 @@ "displayName": "sparsedrive", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-07-20", @@ -13588,7 +13140,7 @@ "functionCalling": false }, "modalities": { - "input": [], + "input": ["video"], "output": ["text"] } }, @@ -13596,7 +13148,6 @@ "displayName": "streampetr", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-11-13", @@ -13607,7 +13158,7 @@ "functionCalling": false }, "modalities": { - "input": [], + "input": ["video"], "output": ["text"] } }, @@ -13615,7 +13166,6 @@ "displayName": "studiovoice", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-06-13", @@ -13634,7 +13184,6 @@ "displayName": "synthetic-video-detector", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "lastUpdated": "2026-04-16", @@ -13645,7 +13194,7 @@ "functionCalling": false }, "modalities": { - "input": [], + "input": ["video"], "output": ["text"] } }, @@ -13653,7 +13202,6 @@ "displayName": "usdcode", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "lastUpdated": "2026-01-01", @@ -13672,7 +13220,6 @@ "displayName": "usdvalidate", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "lastUpdated": "2025-01-08", @@ -13691,7 +13238,6 @@ "displayName": "GPT-OSS-120B", "description": "Open GPT reasoning model for self-hosted agents and controllable deployments", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-08", @@ -13715,7 +13261,6 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": true, @@ -13738,7 +13283,6 @@ "displayName": "Whisper Large v3", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-09", @@ -13758,7 +13302,6 @@ "displayName": "Laguna XS 2.1", "description": "Agentic coding model from Poolside in the XS size class for local deployment", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 16384, "structuredOutput": false, @@ -13778,7 +13321,6 @@ "displayName": "Qwen Image", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 0, "structuredOutput": false, @@ -13798,7 +13340,6 @@ "displayName": "Qwen Image Edit", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 0, "maxOutputTokens": 0, "structuredOutput": false, @@ -13818,7 +13359,6 @@ "displayName": "Qwen2.5 Coder 32b Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 4096, "structuredOutput": true, @@ -13838,7 +13378,6 @@ "displayName": "Qwen3 Coder 480B A35B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 66536, "knowledgeCutoff": "2025-04", @@ -13858,7 +13397,6 @@ "displayName": "Qwen3-Next-80B-A3B-Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-12", @@ -13878,7 +13416,6 @@ "displayName": "Qwen3.5 122B-A10B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -13893,7 +13430,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -13901,7 +13438,6 @@ "displayName": "Qwen3.5-397B-A17B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 262144, "maxOutputTokens": 8192, "knowledgeCutoff": "2026-01", @@ -13925,7 +13461,6 @@ "displayName": "sarvam-m", "description": "Efficient Indian-language reasoning model for chat, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-07-25", @@ -13944,7 +13479,6 @@ "displayName": "Step 3.5 Flash", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 256000, "maxOutputTokens": 16384, "lastUpdated": "2026-02-02", @@ -13966,7 +13500,6 @@ "displayName": "Step 3.7 Flash", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 256000, "maxOutputTokens": 16384, "lastUpdated": "2026-05-28", @@ -13988,7 +13521,6 @@ "displayName": "Inkling", "description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 1048576, "maxOutputTokens": 16384, "lastUpdated": "2026-07-15", @@ -14007,7 +13539,6 @@ "displayName": "solar-10.7b-instruct", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 128000, "maxOutputTokens": 8192, "lastUpdated": "2025-04-10", @@ -14026,7 +13557,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.api.nvidia.com/nim/", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -14051,7 +13581,6 @@ "displayName": "deepseek-v4-flash", "description": "Fast DeepSeek model for efficient chat, coding help, and agent loops", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 1048576, "maxOutputTokens": 1048576, "lastUpdated": "2026-04-24", @@ -14073,7 +13602,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 1048576, "maxOutputTokens": 1048576, "knowledgeCutoff": "2025-05", @@ -14097,7 +13625,6 @@ "displayName": "deepseek-v4-pro", "description": "Flagship DeepSeek model for coding, reasoning, and agentic work", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 1048576, "maxOutputTokens": 1048576, "lastUpdated": "2026-04-24", @@ -14119,7 +13646,6 @@ "displayName": "gemma4:31b", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -14141,7 +13667,6 @@ "displayName": "glm-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 202752, "maxOutputTokens": 131072, "lastUpdated": "2026-04-07", @@ -14162,7 +13687,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 976000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -14180,11 +13704,52 @@ "output": ["text"] } }, + "glm-5.3": { + "displayName": "GLM-5.3", + "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "glm-5.3-flash": { + "displayName": "GLM-5.3-Flash", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "video", "pdf"], + "output": ["text"] + } + }, "gpt-oss:120b": { "displayName": "gpt-oss:120b", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 131072, "maxOutputTokens": 32768, "lastUpdated": "2026-01-19", @@ -14205,7 +13770,6 @@ "displayName": "gpt-oss:20b", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 131072, "maxOutputTokens": 32768, "lastUpdated": "2026-01-19", @@ -14226,7 +13790,6 @@ "displayName": "kimi-k2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 262144, "lastUpdated": "2026-01-27", @@ -14247,7 +13810,6 @@ "displayName": "kimi-k2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 262144, "lastUpdated": "2026-04-20", @@ -14268,7 +13830,6 @@ "displayName": "kimi-k2.7-code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -14291,7 +13852,6 @@ "displayName": "kimi-k3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -14314,7 +13874,6 @@ "displayName": "minimax-m2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -14333,7 +13892,6 @@ "displayName": "minimax-m2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 196608, "maxOutputTokens": 196608, "lastUpdated": "2026-03-18", @@ -14354,7 +13912,6 @@ "displayName": "minimax-m3", "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 512000, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -14369,7 +13926,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -14377,7 +13934,6 @@ "displayName": "mistral-large-3:675b", "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 262144, "lastUpdated": "2026-01-19", @@ -14395,7 +13951,6 @@ "displayName": "nemotron-3-nano:30b", "description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 1048576, "maxOutputTokens": 131072, "lastUpdated": "2026-01-19", @@ -14416,7 +13971,6 @@ "displayName": "nemotron-3-super", "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 65536, "lastUpdated": "2026-03-12", @@ -14437,7 +13991,6 @@ "displayName": "nemotron-3-ultra", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 128000, "lastUpdated": "2026-06-04", @@ -14458,7 +14011,6 @@ "displayName": "qwen3.5:397b", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.ollama.com/cloud", "contextWindow": 262144, "maxOutputTokens": 65536, "lastUpdated": "2026-02-17", @@ -14481,7 +14033,6 @@ "displayName": "chatgpt-image-latest", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 0, "inputLimit": 0, "maxOutputTokens": 0, @@ -14500,7 +14051,6 @@ "displayName": "GPT-3.5-turbo", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 16385, "maxOutputTokens": 4096, "knowledgeCutoff": "2021-09-01", @@ -14520,7 +14070,6 @@ "displayName": "GPT-4", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 8192, "maxOutputTokens": 8192, "knowledgeCutoff": "2023-11", @@ -14540,7 +14089,6 @@ "displayName": "GPT-4 Turbo", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -14560,7 +14108,6 @@ "displayName": "GPT-4.1", "description": "Long-lived GPT workhorse for coding, instruction following, and production apps", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -14580,7 +14127,6 @@ "displayName": "GPT-4.1 mini", "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -14600,7 +14146,6 @@ "displayName": "GPT-4.1 nano", "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -14620,7 +14165,6 @@ "displayName": "GPT-4o", "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -14640,7 +14184,6 @@ "displayName": "GPT-4o (2024-05-13)", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-09", @@ -14660,7 +14203,6 @@ "displayName": "GPT-4o (2024-08-06)", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -14680,7 +14222,6 @@ "displayName": "GPT-4o (2024-11-20)", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -14700,7 +14241,6 @@ "displayName": "GPT-4o mini", "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -14720,7 +14260,6 @@ "displayName": "GPT-5", "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -14744,7 +14283,6 @@ "displayName": "GPT-5 Mini", "description": "Small GPT-5 for responsive agents, coding help, and everyday automation", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -14768,7 +14306,6 @@ "displayName": "GPT-5 Nano", "description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -14792,7 +14329,6 @@ "displayName": "GPT-5 Pro", "description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 272000, @@ -14816,7 +14352,6 @@ "displayName": "GPT-5.1", "description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -14840,7 +14375,6 @@ "displayName": "GPT-5.2", "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -14864,7 +14398,6 @@ "displayName": "GPT-5.2 Chat", "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-08-31", @@ -14887,7 +14420,6 @@ "displayName": "GPT-5.2 Pro", "description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -14911,7 +14443,6 @@ "displayName": "GPT-5.3 Chat (latest)", "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-08-31", @@ -14931,7 +14462,6 @@ "displayName": "GPT-5.3 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -14955,7 +14485,6 @@ "displayName": "GPT-5.3 Codex Spark", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "inputLimit": 100000, "maxOutputTokens": 32000, @@ -14979,7 +14508,6 @@ "displayName": "GPT-5.4", "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15003,7 +14531,6 @@ "displayName": "GPT-5.4 mini", "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -15027,7 +14554,6 @@ "displayName": "GPT-5.4 nano", "description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -15051,7 +14577,6 @@ "displayName": "GPT-5.4 Pro", "description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15075,7 +14600,6 @@ "displayName": "GPT-5.5", "description": "Default frontier GPT for coding, computer use, research, and knowledge work", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15099,7 +14623,6 @@ "displayName": "GPT-5.5 Pro", "description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15123,7 +14646,6 @@ "displayName": "GPT-5.6", "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15147,7 +14669,6 @@ "displayName": "GPT-5.6 Luna", "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15171,7 +14692,6 @@ "displayName": "GPT-5.6 Sol", "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15195,7 +14715,6 @@ "displayName": "GPT-5.6 Terra", "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -15219,7 +14738,6 @@ "displayName": "gpt-image-1", "description": "OpenAI image model for production generation, edits, and brand-safe visual workflows", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 0, "inputLimit": 0, "maxOutputTokens": 0, @@ -15238,7 +14756,6 @@ "displayName": "gpt-image-1-mini", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 0, "inputLimit": 0, "maxOutputTokens": 0, @@ -15257,7 +14774,6 @@ "displayName": "gpt-image-1.5", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 0, "inputLimit": 0, "maxOutputTokens": 0, @@ -15276,7 +14792,6 @@ "displayName": "gpt-image-2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 0, "inputLimit": 0, "maxOutputTokens": 0, @@ -15295,7 +14810,6 @@ "displayName": "GPT-Realtime-2.1", "description": "Realtime speech-to-speech model with configurable reasoning, tool use, and robust voice-agent behavior", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 128000, "inputLimit": 96000, "maxOutputTokens": 32000, @@ -15319,7 +14833,6 @@ "displayName": "o1", "description": "O-series reasoning model for hard analysis, math, coding, and planning", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2023-09", @@ -15342,7 +14855,6 @@ "displayName": "o1-pro", "description": "O-series reasoning model for hard analysis, math, coding, and planning", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2023-09", @@ -15365,7 +14877,6 @@ "displayName": "o3", "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -15388,7 +14899,6 @@ "displayName": "o3-mini", "description": "Smaller o-series reasoner for economical coding, math, and planning tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -15411,7 +14921,6 @@ "displayName": "o3-pro", "description": "High-effort o3 tier for difficult technical reasoning and careful answers", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -15434,7 +14943,6 @@ "displayName": "o4-mini", "description": "Fast o-series model for compact reasoning, coding, and tool use", "lifecycle": "deprecated", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -15457,7 +14965,6 @@ "displayName": "text-embedding-3-large", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 8191, "maxOutputTokens": 3072, "knowledgeCutoff": "2024-01", @@ -15476,7 +14983,6 @@ "displayName": "text-embedding-3-small", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 8191, "maxOutputTokens": 1536, "knowledgeCutoff": "2024-01", @@ -15495,7 +15001,6 @@ "displayName": "text-embedding-ada-002", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://platform.openai.com/docs/models", "contextWindow": 8192, "maxOutputTokens": 1536, "knowledgeCutoff": "2022-12", @@ -15516,7 +15021,6 @@ "displayName": "Big Pickle", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "inputLimit": 160000, "maxOutputTokens": 32000, @@ -15538,7 +15042,6 @@ "displayName": "Claude Haiku 3.5", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-07-31", @@ -15557,7 +15060,6 @@ "displayName": "Claude Fable 5", "description": "Claude model for creative writing, analysis, and controlled agent workflows", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -15579,7 +15081,6 @@ "displayName": "Claude Haiku 4.5", "description": "Fast Claude model for responsive assistance, classification, and lightweight agents", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-02-28", @@ -15598,7 +15099,6 @@ "displayName": "Claude Opus 4.1", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-03-31", @@ -15617,7 +15117,6 @@ "displayName": "Claude Opus 4.5", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-03-31", @@ -15639,7 +15138,6 @@ "displayName": "Claude Opus 4.6", "description": "High-end Claude for difficult coding, planning, and slower expert reasoning", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05-31", @@ -15661,7 +15159,6 @@ "displayName": "Claude Opus 4.7", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -15683,7 +15180,6 @@ "displayName": "Claude Opus 4.8", "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01", @@ -15705,7 +15201,6 @@ "displayName": "Claude Opus 5", "description": "Strongest Claude Opus model for coding, agents, and professional work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-05", @@ -15727,7 +15222,6 @@ "displayName": "Claude Sonnet 4", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-03-31", @@ -15746,7 +15240,6 @@ "displayName": "Claude Sonnet 4.5", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-07-31", @@ -15765,7 +15258,6 @@ "displayName": "Claude Sonnet 4.6", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-08-31", @@ -15787,7 +15279,6 @@ "displayName": "Claude Sonnet 5", "description": "Everyday Claude agent model for coding, planning, browsing, and general work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -15809,7 +15300,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -15833,7 +15323,6 @@ "displayName": "DeepSeek V4 Flash Free", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05", @@ -15857,7 +15346,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -15881,7 +15369,6 @@ "displayName": "Gemini 3 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -15896,7 +15383,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -15904,7 +15391,6 @@ "displayName": "Gemini 3 Pro", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -15919,7 +15405,7 @@ "efforts": ["low", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -15927,7 +15413,6 @@ "displayName": "Gemini 3.1 Pro Preview", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -15942,7 +15427,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -15950,7 +15435,6 @@ "displayName": "Gemini 3.5 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -15965,7 +15449,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -15973,7 +15457,6 @@ "displayName": "Gemini 3.5 Flash Lite", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -15988,7 +15471,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -15996,7 +15479,6 @@ "displayName": "Gemini 3.6 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -16011,7 +15493,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -16019,7 +15501,6 @@ "displayName": "Gemini 3.7 Flash", "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -16034,7 +15515,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -16042,7 +15523,6 @@ "displayName": "GLM-4.6", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -16064,7 +15544,6 @@ "displayName": "GLM-4.7", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -16086,7 +15565,6 @@ "displayName": "GLM-4.7 Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -16109,7 +15587,6 @@ "displayName": "GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -16131,7 +15608,6 @@ "displayName": "GLM-5 Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -16154,7 +15630,6 @@ "displayName": "GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -16176,7 +15651,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -16198,7 +15672,6 @@ "displayName": "GPT-5", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16222,7 +15695,6 @@ "displayName": "GPT-5 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16246,7 +15718,6 @@ "displayName": "GPT-5 Nano", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16270,7 +15741,6 @@ "displayName": "GPT-5.1", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16294,7 +15764,6 @@ "displayName": "GPT-5.1 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16318,7 +15787,6 @@ "displayName": "GPT-5.1 Codex Max", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16342,7 +15810,6 @@ "displayName": "GPT-5.1 Codex Mini", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16366,7 +15833,6 @@ "displayName": "GPT-5.2", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16390,7 +15856,6 @@ "displayName": "GPT-5.2 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16414,7 +15879,6 @@ "displayName": "GPT-5.3 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16438,7 +15902,6 @@ "displayName": "GPT-5.3 Codex Spark", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 128000, "inputLimit": 128000, "maxOutputTokens": 128000, @@ -16462,7 +15925,6 @@ "displayName": "GPT-5.4", "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -16486,7 +15948,6 @@ "displayName": "GPT-5.4 Mini", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16510,7 +15971,6 @@ "displayName": "GPT-5.4 Nano", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -16534,7 +15994,6 @@ "displayName": "GPT-5.4 Pro", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -16558,7 +16017,6 @@ "displayName": "GPT-5.5", "description": "Default frontier GPT for coding, computer use, research, and knowledge work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -16582,7 +16040,6 @@ "displayName": "GPT-5.5 Pro", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -16606,7 +16063,6 @@ "displayName": "GPT-5.6 Luna", "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -16630,7 +16086,6 @@ "displayName": "GPT-5.6 Sol (50% Off)", "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -16654,7 +16109,6 @@ "displayName": "GPT-5.6 Terra", "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -16678,7 +16132,6 @@ "displayName": "Grok 4.5", "description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 500000, "maxOutputTokens": 500000, "structuredOutput": true, @@ -16700,7 +16153,6 @@ "displayName": "Grok 4.6", "description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 500000, "maxOutputTokens": 500000, "knowledgeCutoff": "2026-02-01", @@ -16723,7 +16175,6 @@ "displayName": "Grok Build 0.1", "description": "Fast Grok coding model tuned for agentic engineering and iterative edits", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 256000, "maxOutputTokens": 256000, "structuredOutput": true, @@ -16742,7 +16193,6 @@ "displayName": "Grok Code Fast 1", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 256000, "maxOutputTokens": 256000, "lastUpdated": "2025-08-20", @@ -16760,8 +16210,7 @@ "hy3-free": { "displayName": "Hy3 Free", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", - "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", + "lifecycle": "deprecated", "contextWindow": 190000, "maxOutputTokens": 64000, "structuredOutput": true, @@ -16785,7 +16234,6 @@ "displayName": "Hy3 preview Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-06", @@ -16805,7 +16253,6 @@ "displayName": "Kimi K2", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-10", @@ -16824,7 +16271,6 @@ "displayName": "Kimi K2 Thinking", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-10", @@ -16843,7 +16289,6 @@ "displayName": "Kimi K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-10", @@ -16857,7 +16302,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -16865,7 +16310,6 @@ "displayName": "Kimi K2.5 Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2024-10", @@ -16880,7 +16324,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -16888,7 +16332,6 @@ "displayName": "Kimi K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-10", @@ -16902,7 +16345,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -16910,7 +16353,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -16922,7 +16364,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -16930,7 +16372,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -16944,7 +16385,7 @@ "efforts": ["max"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -16952,7 +16393,6 @@ "displayName": "Laguna S 2.1 Free", "description": "Agentic coding model from Poolside in the XS size class for local deployment", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 256000, "maxOutputTokens": 32000, "structuredOutput": false, @@ -16975,7 +16415,6 @@ "displayName": "Ling 2.6 Flash Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262100, "maxOutputTokens": 32800, "knowledgeCutoff": "2025-06", @@ -16991,11 +16430,32 @@ "output": ["text"] } }, + "ling-3.0-flash-fin-free": { + "displayName": "Ling 3.0 Flash Fin Free", + "description": "Finance-enhanced model for financial research, multi-step investment workflows, and long-horizon planning and execution", + "lifecycle": "active", + "contextWindow": 262144, + "maxOutputTokens": 32768, + "structuredOutput": false, + "lastUpdated": "2026-08-27", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "ling-3.0-flash-free": { "displayName": "Ling-3.0-flash Free", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -17018,7 +16478,6 @@ "displayName": "Ling-3.0-tiny Free", "description": "Compact MoE model for responsive agents, instruction following, and multi-turn conversations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -17038,7 +16497,6 @@ "displayName": "LongCat-2.0 Free", "description": "Meituan LongCat-2.0, a reasoning model with tool calling and a 1M-token context window", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 131072, "lastUpdated": "2026-06-30", @@ -17060,7 +16518,6 @@ "displayName": "MiMo V2 Flash Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-12", @@ -17080,7 +16537,6 @@ "displayName": "MiMo V2 Omni Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 64000, "knowledgeCutoff": "2024-12", @@ -17100,7 +16556,6 @@ "displayName": "MiMo V2 Pro Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 64000, "knowledgeCutoff": "2024-12", @@ -17120,7 +16575,6 @@ "displayName": "MiMo V2.5 Free", "description": "MiMo omni model for text, image, video, audio, and agents", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "maxOutputTokens": 32000, "knowledgeCutoff": "2024-12", @@ -17132,7 +16586,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -17140,7 +16594,6 @@ "displayName": "MiniMax-M2.1", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -17159,7 +16612,6 @@ "displayName": "MiniMax-M2.1 Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -17179,7 +16631,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -17198,7 +16649,6 @@ "displayName": "MiniMax-M2.5 Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -17218,7 +16668,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -17237,7 +16686,6 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 512000, "maxOutputTokens": 128000, "lastUpdated": "2026-06-01", @@ -17247,7 +16695,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17255,7 +16703,6 @@ "displayName": "MiniMax-M3 Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 200000, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-01", @@ -17270,7 +16717,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17278,7 +16725,6 @@ "displayName": "Muse Spark 1.2", "description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -17292,7 +16738,7 @@ "efforts": ["minimal", "low", "medium", "high", "xhigh"] }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "video", "pdf", "audio"], "output": ["text"] } }, @@ -17300,7 +16746,6 @@ "displayName": "Muse Spark 1.2 Free", "description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -17315,7 +16760,7 @@ "efforts": ["minimal", "low", "medium", "high", "xhigh"] }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "video", "pdf", "audio"], "output": ["text"] } }, @@ -17323,7 +16768,6 @@ "displayName": "Nemotron 3 Super Free", "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-02", @@ -17343,7 +16787,6 @@ "displayName": "Nemotron 3 Ultra Free", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-02", @@ -17363,7 +16806,6 @@ "displayName": "Nemotron 3.5 Lightning Free", "description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -17383,7 +16825,6 @@ "displayName": "North Mini Code Free", "description": "Cohere coding model for practical software engineering and agentic edits", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-09-23", @@ -17407,7 +16848,6 @@ "displayName": "Qwen3 Coder", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -17426,7 +16866,6 @@ "displayName": "Qwen3.5 Plus", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -17440,7 +16879,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17448,7 +16887,6 @@ "displayName": "Qwen3.6 Plus", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -17462,7 +16900,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17470,7 +16908,6 @@ "displayName": "Qwen3.6 Plus Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -17485,7 +16922,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17493,7 +16930,6 @@ "displayName": "Ring 2.6 1T Free", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262000, "maxOutputTokens": 66000, "knowledgeCutoff": "2025-06", @@ -17513,7 +16949,6 @@ "displayName": "Trinity Large Preview", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-06", @@ -17532,8 +16967,7 @@ "x-preview-f-free": { "displayName": "Ox Alpha Free (Unlimited)", "description": "Stealth reasoning model for coding, agentic tasks, and tool use", - "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", + "lifecycle": "deprecated", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -17548,7 +16982,7 @@ "efforts": ["low", "high", "max"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -17558,7 +16992,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -17577,11 +17010,32 @@ "output": ["text"] } }, + "deepseek-v4-flash-vision-exp": { + "displayName": "DeepSeek V4 Flash Vision Exp", + "description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 384000, + "structuredOutput": true, + "lastUpdated": "2026-08-21", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"], + "toggle": true + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "deepseek-v4-pro": { "displayName": "DeepSeek V4 Pro (New)", "description": "Flagship DeepSeek model for coding, reasoning, and agentic work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -17604,7 +17058,6 @@ "displayName": "GLM-5", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 202752, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -17623,7 +17076,6 @@ "displayName": "GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 202752, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -17642,7 +17094,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -17664,7 +17115,6 @@ "displayName": "GLM-5.3", "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -17682,11 +17132,31 @@ "output": ["text"] } }, + "glm-5.3-flash": { + "displayName": "GLM-5.3-Flash (2x usage)", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "video", "pdf"], + "output": ["text"] + } + }, "gpt-5.6-luna": { "displayName": "GPT-5.6 Luna", "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -17709,8 +17179,7 @@ "grok-4.5": { "displayName": "Grok 4.5", "description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk", - "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", + "lifecycle": "deprecated", "contextWindow": 500000, "maxOutputTokens": 500000, "structuredOutput": true, @@ -17728,11 +17197,32 @@ "output": ["text"] } }, + "grok-4.6": { + "displayName": "Grok 4.6", + "description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects", + "lifecycle": "active", + "contextWindow": 500000, + "maxOutputTokens": 500000, + "knowledgeCutoff": "2026-02-01", + "structuredOutput": true, + "lastUpdated": "2026-08-12", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "high", "xhigh"] + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "hy3": { - "displayName": "Hy3 (8x usage)", + "displayName": "Hy3", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 256000, "maxOutputTokens": 64000, "lastUpdated": "2026-07-06", @@ -17749,11 +17239,30 @@ "output": ["text"] } }, + "hy4-preview": { + "displayName": "Hy4 preview", + "description": "A next-generation productivity model with significantly enhanced Agent and complex task execution capabilities.", + "lifecycle": "active", + "contextWindow": 1024000, + "maxOutputTokens": 64000, + "lastUpdated": "2026-08-28", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "high"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "kimi-k2.5": { "displayName": "Kimi K2.5", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-10", @@ -17764,7 +17273,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17772,7 +17281,6 @@ "displayName": "Kimi K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-10", @@ -17783,7 +17291,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17791,7 +17299,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -17803,7 +17310,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17811,7 +17318,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -17825,7 +17331,27 @@ "efforts": ["max"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "longcat-2.0": { + "displayName": "LongCat-2.0", + "description": "Meituan LongCat-2.0, a reasoning model with tool calling and a 1M-token context window", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "lastUpdated": "2026-06-30", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text"], "output": ["text"] } }, @@ -17833,7 +17359,6 @@ "displayName": "MiMo V2 Omni", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-12", @@ -17852,7 +17377,6 @@ "displayName": "MiMo V2 Pro", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-12", @@ -17871,7 +17395,6 @@ "displayName": "MiMo V2.5", "description": "MiMo omni model for text, image, video, audio, and agents", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-12", @@ -17882,7 +17405,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -17890,7 +17413,6 @@ "displayName": "MiMo V2.5 Pro", "description": "MiMo pro model for strong multimodal reasoning and agent execution", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-12", @@ -17909,7 +17431,6 @@ "displayName": "MiniMax-M2.5", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -17928,7 +17449,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -17947,7 +17467,6 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -17961,7 +17480,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -17969,7 +17488,6 @@ "displayName": "Muse Spark 1.2 Contributor", "description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -17983,15 +17501,14 @@ "efforts": ["minimal", "low", "medium", "high", "xhigh"] }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "video", "pdf", "audio"], "output": ["text"] } }, "ox-alpha-free": { "displayName": "Ox Alpha Free (Unlimited)", "description": "Stealth reasoning model for coding, agentic tasks, and tool use", - "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", + "lifecycle": "deprecated", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -18006,7 +17523,7 @@ "efforts": ["low", "high", "max"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -18014,7 +17531,6 @@ "displayName": "Qwen3.5 Plus", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -18028,7 +17544,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -18036,7 +17552,6 @@ "displayName": "Qwen3.6 Plus", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -18050,7 +17565,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -18058,7 +17573,6 @@ "displayName": "Qwen3.7 Max", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 65536, "lastUpdated": "2026-05-21", @@ -18079,7 +17593,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 65536, "lastUpdated": "2026-06-02", @@ -18092,7 +17605,29 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "qwen3.8-flash": { + "displayName": "Qwen3.8 Flash", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"], + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -18100,7 +17635,6 @@ "displayName": "Qwen3.8 Max", "description": "2.4-trillion-parameter multimodal flagship for coding, professional work, and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://opencode.ai/docs/zen", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -18111,10 +17645,11 @@ "functionCalling": true }, "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"], "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -18124,7 +17659,6 @@ "displayName": "Claude Fable Latest", "description": "Claude model for creative writing, analysis, and controlled agent workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -18146,7 +17680,6 @@ "displayName": "Anthropic Claude Haiku Latest", "description": "Fast Claude model for responsive assistance, classification, and lightweight agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 64000, "structuredOutput": true, @@ -18168,7 +17701,6 @@ "displayName": "Claude Opus Latest", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -18191,7 +17723,6 @@ "displayName": "Anthropic Claude Sonnet Latest", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -18215,9 +17746,8 @@ "displayName": "DeepSeek V4 Flash Latest", "description": "Fast DeepSeek model for efficient chat, coding help, and agent loops", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1310720, - "maxOutputTokens": 1048576, + "maxOutputTokens": 393216, "structuredOutput": true, "lastUpdated": "2026-08-01", "capabilities": { @@ -18238,7 +17768,6 @@ "displayName": "Google Gemini Flash Latest", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01-01", @@ -18253,7 +17782,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "video", "pdf", "audio"], "output": ["text"] } }, @@ -18261,7 +17790,6 @@ "displayName": "Google Gemini Pro Latest", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -18276,7 +17804,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["audio", "pdf", "image", "text"], + "input": ["audio", "pdf", "image", "text", "video"], "output": ["text"] } }, @@ -18284,9 +17812,8 @@ "displayName": "MoonshotAI Kimi Latest", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, - "maxOutputTokens": 974842, + "maxOutputTokens": 943718, "structuredOutput": true, "lastUpdated": "2026-04-27", "capabilities": { @@ -18299,7 +17826,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -18307,7 +17834,6 @@ "displayName": "OpenAI GPT Latest", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-02-16", @@ -18330,7 +17856,6 @@ "displayName": "OpenAI GPT Mini Latest", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -18353,9 +17878,8 @@ "displayName": "Grok Latest", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 500000, - "maxOutputTokens": 1000000, + "maxOutputTokens": 450000, "structuredOutput": true, "lastUpdated": "2026-07-08", "capabilities": { @@ -18375,10 +17899,9 @@ "displayName": "GLM Latest", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 1048576, - "maxOutputTokens": 131072, - "structuredOutput": false, + "contextWindow": 1310720, + "maxOutputTokens": 943718, + "structuredOutput": true, "lastUpdated": "2026-08-19", "capabilities": { "vision": false, @@ -18397,7 +17920,6 @@ "displayName": "Aion-2.0", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": false, @@ -18416,7 +17938,6 @@ "displayName": "Aion-3.0", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": false, @@ -18435,7 +17956,6 @@ "displayName": "Aion-3.0-Mini", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": false, @@ -18454,9 +17974,8 @@ "displayName": "Aion-RP 1.0 (8B)", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, - "maxOutputTokens": 32768, + "maxOutputTokens": 29491, "knowledgeCutoff": "2023-12-31", "structuredOutput": false, "lastUpdated": "2025-02-04", @@ -18470,30 +17989,10 @@ "output": ["text"] } }, - "allenai/olmo-3-32b-think": { - "displayName": "Olmo 3 32B Think", - "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 65536, - "maxOutputTokens": 65536, - "structuredOutput": true, - "lastUpdated": "2025-11-21", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": false - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "amazon/nova-2-lite-v1": { "displayName": "Nova 2 Lite", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65535, "structuredOutput": false, @@ -18503,8 +18002,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } }, @@ -18512,7 +18014,6 @@ "displayName": "Nova Lite 1.0", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 300000, "maxOutputTokens": 5120, "knowledgeCutoff": "2024-10-31", @@ -18532,7 +18033,6 @@ "displayName": "Nova Micro 1.0", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 5120, "knowledgeCutoff": "2024-10-31", @@ -18552,7 +18052,6 @@ "displayName": "Nova Premier 1.0", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 32000, "structuredOutput": false, @@ -18571,7 +18070,6 @@ "displayName": "Nova Pro 1.0", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 300000, "maxOutputTokens": 5120, "knowledgeCutoff": "2024-10-31", @@ -18591,7 +18089,6 @@ "displayName": "Magnum v4 72B", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-06-30", @@ -18611,7 +18108,6 @@ "displayName": "Claude 3 Haiku", "description": "Fast Claude model for responsive assistance, classification, and lightweight agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-08-31", @@ -18631,7 +18127,6 @@ "displayName": "Claude Fable 5", "description": "Claude model for creative writing, analysis, and controlled agent workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -18654,7 +18149,6 @@ "displayName": "Claude Haiku 4.5 (latest)", "description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-02-28", @@ -18677,7 +18171,6 @@ "displayName": "Claude Opus 4", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-01-31", @@ -18700,7 +18193,6 @@ "displayName": "Claude Opus 4.1 (latest)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-03-31", @@ -18723,7 +18215,6 @@ "displayName": "Claude Opus 4.5 (latest)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-05", @@ -18735,7 +18226,6 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"], "toggle": true }, "modalities": { @@ -18747,7 +18237,6 @@ "displayName": "Claude Opus 4.6", "description": "High-end Claude for difficult coding, planning, and slower expert reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05-31", @@ -18771,7 +18260,6 @@ "displayName": "Claude Opus 4.7", "description": "Stronger Opus tier for advanced software work and high-stakes reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -18795,7 +18283,6 @@ "displayName": "Claude Opus 4.7 (Fast)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -18819,7 +18306,6 @@ "displayName": "Claude Opus 4.8", "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01", @@ -18843,7 +18329,6 @@ "displayName": "Claude Opus 4.8 (Fast)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01", @@ -18867,7 +18352,6 @@ "displayName": "Claude Opus 5", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-05", @@ -18891,7 +18375,6 @@ "displayName": "Claude Opus 5 (Fast)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-05", @@ -18915,7 +18398,6 @@ "displayName": "Claude Sonnet 4", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-31", @@ -18938,7 +18420,6 @@ "displayName": "Claude Sonnet 4.5 (latest)", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-07-31", @@ -18961,7 +18442,6 @@ "displayName": "Claude Sonnet 4.6", "description": "Claude workhorse for coding agents, careful analysis, and production cost control", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -18985,7 +18465,6 @@ "displayName": "Claude Sonnet 5", "description": "Everyday Claude agent model for coding, planning, browsing, and general work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -19009,10 +18488,9 @@ "displayName": "Trinity Large Thinking", "description": "Reasoning-optimized 398B MoE agent model with extended thinking for long-horizon and multi-turn tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, - "structuredOutput": true, + "maxOutputTokens": 80000, + "structuredOutput": false, "lastUpdated": "2026-05-28", "capabilities": { "vision": false, @@ -19024,31 +18502,10 @@ "output": ["text"] } }, - "arcee-ai/virtuoso-large": { - "displayName": "Virtuoso Large", - "description": "Flagship model for demanding analysis, coding, and production agent workflows", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 131072, - "maxOutputTokens": 64000, - "knowledgeCutoff": "2025-03-31", - "structuredOutput": false, - "lastUpdated": "2025-05-05", - "capabilities": { - "vision": false, - "reasoning": false, - "functionCalling": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "baidu/ernie-4.5-vl-424b-a47b": { "displayName": "ERNIE 4.5 VL 424B A47B ", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 123000, "maxOutputTokens": 16000, "knowledgeCutoff": "2025-03-31", @@ -19059,6 +18516,9 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["image", "text"], "output": ["text"] @@ -19068,7 +18528,6 @@ "displayName": "Seed 1.6", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -19078,8 +18537,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } }, @@ -19087,7 +18549,6 @@ "displayName": "Seed 1.6 Flash", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -19097,8 +18558,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } }, @@ -19106,9 +18570,8 @@ "displayName": "Seed 2.1 Turbo", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-08-12", "capabilities": { @@ -19116,8 +18579,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -19125,7 +18591,6 @@ "displayName": "Seed 2.0 Code", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -19140,7 +18605,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -19148,7 +18613,6 @@ "displayName": "Seed 2.0 Lite", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -19163,7 +18627,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -19171,7 +18635,6 @@ "displayName": "Seed 2.0 Mini", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -19186,7 +18649,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -19194,7 +18657,6 @@ "displayName": "UI-TARS 7B ", "description": "Multimodal model for analyzing text, images, documents, and rich media", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 2048, "knowledgeCutoff": "2025-01-31", @@ -19214,7 +18676,6 @@ "displayName": "Uncensored", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-04-30", @@ -19234,7 +18695,6 @@ "displayName": "Command A", "description": "Cohere command model for multilingual enterprise agents, tools, and chat", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-08-31", @@ -19254,7 +18714,6 @@ "displayName": "Command R", "description": "Cohere retrieval model for long-context chat and enterprise RAG workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 4000, "knowledgeCutoff": "2024-06-01", @@ -19274,7 +18733,6 @@ "displayName": "Command R+", "description": "Cohere's RAG workhorse for long-context enterprise search and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 4000, "knowledgeCutoff": "2024-06-01", @@ -19294,7 +18752,6 @@ "displayName": "Command R7B", "description": "Cohere retrieval model for long-context chat and enterprise RAG workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 4000, "knowledgeCutoff": "2024-06-01", @@ -19314,7 +18771,6 @@ "displayName": "North Mini Code (free)", "description": "Cohere coding model for practical software engineering and agentic edits", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, "maxOutputTokens": 64000, "structuredOutput": false, @@ -19325,24 +18781,8 @@ "reasoning": true, "functionCalling": true }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, - "deepcogito/cogito-v2.1-671b": { - "displayName": "Cogito v2.1 671B", - "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 128000, - "maxOutputTokens": 128000, - "structuredOutput": true, - "lastUpdated": "2025-11-13", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": false + "thinkingOptions": { + "toggle": true }, "modalities": { "input": ["text"], @@ -19353,7 +18793,6 @@ "displayName": "DeepSeek Chat", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 163840, "maxOutputTokens": 16000, "knowledgeCutoff": "2025-09", @@ -19373,9 +18812,8 @@ "displayName": "DeepSeek V3 0324", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 163840, - "maxOutputTokens": 163840, + "maxOutputTokens": 147456, "knowledgeCutoff": "2024-07-31", "structuredOutput": true, "lastUpdated": "2025-03-24", @@ -19393,9 +18831,8 @@ "displayName": "DeepSeek V3.1", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 163840, - "maxOutputTokens": 32768, + "maxOutputTokens": 144900, "knowledgeCutoff": "2025-03-31", "structuredOutput": true, "lastUpdated": "2025-08-21", @@ -19416,7 +18853,6 @@ "displayName": "DeepSeek-R1", "description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 64000, "maxOutputTokens": 16000, "knowledgeCutoff": "2024-07", @@ -19436,7 +18872,6 @@ "displayName": "R1 0528", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 163840, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -19456,9 +18891,8 @@ "displayName": "R1 Distill Llama 70B", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8192, - "maxOutputTokens": 8192, + "maxOutputTokens": 7372, "knowledgeCutoff": "2024-07-31", "structuredOutput": false, "lastUpdated": "2025-01-23", @@ -19467,6 +18901,9 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -19476,7 +18913,6 @@ "displayName": "DeepSeek V3.1 Terminus", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 163840, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -19499,7 +18935,6 @@ "displayName": "DeepSeek V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 163840, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-07", @@ -19522,7 +18957,6 @@ "displayName": "DeepSeek V3.2 Exp", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 163840, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-07-31", @@ -19545,7 +18979,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -19569,9 +19002,8 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1310720, - "maxOutputTokens": 384000, + "maxOutputTokens": 943718, "knowledgeCutoff": "2025-05", "structuredOutput": true, "lastUpdated": "2026-07-31", @@ -19593,7 +19025,6 @@ "displayName": "DeepSeek V4 Flash Vision Exp", "description": "Fast DeepSeek model for efficient chat, coding help, and agent loops", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 384000, "structuredOutput": false, @@ -19616,7 +19047,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 393216, "knowledgeCutoff": "2025-05", @@ -19640,11 +19070,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -19663,9 +19092,8 @@ "displayName": "Dots3-Note Preview (free)", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 512000, - "maxOutputTokens": 512000, + "maxOutputTokens": 460800, "structuredOutput": true, "lastUpdated": "2026-08-14", "isFree": true, @@ -19674,6 +19102,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -19683,7 +19114,6 @@ "displayName": "Gemini 2.5 Flash", "description": "Fast Gemini workhorse for multimodal apps where latency and price matter", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65535, "knowledgeCutoff": "2025-01", @@ -19698,7 +19128,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -19706,7 +19136,6 @@ "displayName": "Nano Banana", "description": "Nano Banana image model for fast generation, edits, and character-consistent assets", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-06", @@ -19726,7 +19155,6 @@ "displayName": "Gemini 2.5 Flash-Lite", "description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65535, "knowledgeCutoff": "2025-01", @@ -19741,7 +19169,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -19749,7 +19177,6 @@ "displayName": "Gemini 2.5 Pro", "description": "Google's proven reasoning model for coding, math, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -19761,7 +19188,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -19769,7 +19196,6 @@ "displayName": "Gemini 2.5 Pro Preview 06-05", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01-31", @@ -19789,7 +19215,6 @@ "displayName": "Gemini 2.5 Pro Preview 05-06", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65535, "knowledgeCutoff": "2025-01-31", @@ -19801,7 +19226,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "pdf", "audio", "video"], "output": ["text"] } }, @@ -19809,7 +19234,6 @@ "displayName": "Gemini 3 Flash Preview", "description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -19825,7 +19249,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -19833,7 +19257,6 @@ "displayName": "Nano Banana Pro", "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -19853,7 +19276,6 @@ "displayName": "Nano Banana Pro", "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -19873,7 +19295,6 @@ "displayName": "Nano Banana 2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -19897,9 +19318,8 @@ "displayName": "Nano Banana 2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, - "maxOutputTokens": 65536, + "maxOutputTokens": 58982, "knowledgeCutoff": "2025-01", "structuredOutput": true, "lastUpdated": "2026-02-26", @@ -19921,7 +19341,6 @@ "displayName": "Gemini 3.1 Flash Lite", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -19937,7 +19356,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -19945,9 +19364,8 @@ "displayName": "Nano Banana 2 Lite", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, - "maxOutputTokens": 65536, + "maxOutputTokens": 58982, "knowledgeCutoff": "2025-01", "structuredOutput": false, "lastUpdated": "2026-06-30", @@ -19969,7 +19387,6 @@ "displayName": "Gemini 3.1 Flash Lite Preview", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -19985,7 +19402,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -19993,7 +19410,6 @@ "displayName": "Gemini 3.1 Pro Preview", "description": "Reasoning-first Gemini preview for agentic coding and complex problem solving", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -20008,7 +19424,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -20016,7 +19432,6 @@ "displayName": "Gemini 3.1 Pro Preview Custom Tools", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -20031,7 +19446,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -20039,7 +19454,6 @@ "displayName": "Gemini 3.5 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -20054,7 +19468,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -20062,7 +19476,6 @@ "displayName": "Gemini 3.5 Flash Lite", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -20077,7 +19490,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -20085,7 +19498,6 @@ "displayName": "Gemini 3.6 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -20100,7 +19512,7 @@ "efforts": ["minimal", "low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -20108,7 +19520,6 @@ "displayName": "Gemini 3.7 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -20123,7 +19534,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -20131,7 +19542,6 @@ "displayName": "Gemma 2 27B", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8192, "maxOutputTokens": 2048, "knowledgeCutoff": "2024-06-30", @@ -20151,7 +19561,6 @@ "displayName": "Gemma 3 12B", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-08-31", @@ -20171,9 +19580,8 @@ "displayName": "Gemma 3 27B", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 262144, - "maxOutputTokens": 131072, + "contextWindow": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2024-08-31", "structuredOutput": true, "lastUpdated": "2025-03-12", @@ -20191,7 +19599,6 @@ "displayName": "Gemma 3 4B", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-08-31", @@ -20207,31 +19614,10 @@ "output": ["text"] } }, - "google/gemma-3n-e4b-it": { - "displayName": "Gemma 3n 4B", - "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 32768, - "maxOutputTokens": 32768, - "knowledgeCutoff": "2024-08-31", - "structuredOutput": true, - "lastUpdated": "2025-05-20", - "capabilities": { - "vision": false, - "reasoning": false, - "functionCalling": false - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "google/gemma-4-26b-a4b-it": { "displayName": "Gemma 4 26B A4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 16384, "structuredOutput": true, @@ -20245,7 +19631,7 @@ "toggle": true }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } }, @@ -20253,7 +19639,6 @@ "displayName": "Gemma 4 26B A4B (free)", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -20268,7 +19653,7 @@ "toggle": true }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } }, @@ -20276,9 +19661,8 @@ "displayName": "Gemma 4 31B IT", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 16384, "structuredOutput": true, "lastUpdated": "2026-04-02", "capabilities": { @@ -20290,7 +19674,7 @@ "toggle": true }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } }, @@ -20298,7 +19682,6 @@ "displayName": "Gemma 4 31B (free)", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -20313,7 +19696,7 @@ "toggle": true }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } }, @@ -20321,7 +19704,6 @@ "displayName": "Lyria 3 Clip Preview", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "structuredOutput": false, @@ -20341,7 +19723,6 @@ "displayName": "Lyria 3 Pro Preview", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 65536, "structuredOutput": false, @@ -20361,9 +19742,8 @@ "displayName": "MythoMax 13B", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8192, - "maxOutputTokens": 4096, + "maxOutputTokens": 3686, "knowledgeCutoff": "2023-06-30", "structuredOutput": true, "lastUpdated": "2023-07-02", @@ -20381,9 +19761,8 @@ "displayName": "Granite 4.0 Micro", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131000, - "maxOutputTokens": 131000, + "maxOutputTokens": 117900, "structuredOutput": false, "lastUpdated": "2025-10-20", "capabilities": { @@ -20400,9 +19779,8 @@ "displayName": "Granite 4.1 8B", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "structuredOutput": true, "lastUpdated": "2026-04-30", "capabilities": { @@ -20415,60 +19793,42 @@ "output": ["text"] } }, - "inception/mercury-2": { - "displayName": "Mercury 2", + "ibm-granite/granite-4.2-8b": { + "displayName": "Granite 4.2 8B", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 128000, - "maxOutputTokens": 50000, + "contextWindow": 131072, + "maxOutputTokens": 117964, "structuredOutput": true, - "lastUpdated": "2026-03-04", + "lastUpdated": "2026-08-31", "capabilities": { "vision": false, "reasoning": true, "functionCalling": true }, "thinkingOptions": { - "efforts": ["none", "low", "medium", "high"] + "efforts": ["none", "low", "high"] }, "modalities": { "input": ["text"], "output": ["text"] } }, - "inclusionai/ling-2.6-1t": { - "displayName": "Ling-2.6-1T", - "description": "Tool-capable chat model for instruction following and agentic application workflows", + "inception/mercury-2": { + "displayName": "Mercury 2", + "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 262144, - "maxOutputTokens": 32768, + "contextWindow": 128000, + "maxOutputTokens": 50000, "structuredOutput": true, - "lastUpdated": "2026-04-23", + "lastUpdated": "2026-03-04", "capabilities": { "vision": false, - "reasoning": false, + "reasoning": true, "functionCalling": true }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, - "inclusionai/ling-2.6-flash": { - "displayName": "Ling-2.6-flash", - "description": "Efficient model for low-latency assistance, extraction, and routine automation", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 262144, - "maxOutputTokens": 32768, - "structuredOutput": true, - "lastUpdated": "2026-04-21", - "capabilities": { - "vision": false, - "reasoning": false, - "functionCalling": true + "thinkingOptions": { + "efforts": ["none", "low", "medium", "high"] }, "modalities": { "input": ["text"], @@ -20479,7 +19839,6 @@ "displayName": "Ling-3.0-flash", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -20489,46 +19848,30 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] } }, - "inclusionai/ring-2.6-1t": { - "displayName": "Ring-2.6-1T", - "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", + "inclusionai/ling-3.0-flash-fin:free": { + "displayName": "Ling 3.0 Flash Fin (free)", + "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 65536, + "maxOutputTokens": 32768, "structuredOutput": false, - "lastUpdated": "2026-05-08", + "lastUpdated": "2026-08-27", + "isFree": true, "capabilities": { "vision": false, "reasoning": true, "functionCalling": true }, "thinkingOptions": { - "efforts": ["high", "xhigh"] - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, - "kwaipilot/kat-coder-air-v2.5": { - "displayName": "KAT-Coder-Air V2.5", - "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 256000, - "maxOutputTokens": 80000, - "structuredOutput": true, - "lastUpdated": "2026-07-10", - "capabilities": { - "vision": false, - "reasoning": false, - "functionCalling": true + "toggle": true }, "modalities": { "input": ["text"], @@ -20539,9 +19882,8 @@ "displayName": "KAT-Coder-Pro V2", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 80000, + "maxOutputTokens": 144000, "structuredOutput": true, "lastUpdated": "2026-03-27", "capabilities": { @@ -20558,9 +19900,8 @@ "displayName": "KAT-Coder-Pro V2.5", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 256000, - "maxOutputTokens": 80000, + "contextWindow": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-07-10", "capabilities": { @@ -20577,7 +19918,6 @@ "displayName": "LFM2.5-2.6B (free)", "description": "Free provider route for experiments, demos, and cost-sensitive chat workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, "maxOutputTokens": 8192, "structuredOutput": true, @@ -20597,7 +19937,6 @@ "displayName": "Weaver (alpha)", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8000, "maxOutputTokens": 6000, "knowledgeCutoff": "2023-06-30", @@ -20617,7 +19956,6 @@ "displayName": "LongCat 2.0", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048756, "maxOutputTokens": 262144, "structuredOutput": false, @@ -20627,6 +19965,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -20636,7 +19977,6 @@ "displayName": "Llama 3.1 70B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-12-31", @@ -20656,9 +19996,8 @@ "displayName": "Llama-3.1-8B-Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2023-12", "structuredOutput": true, "lastUpdated": "2024-07-23", @@ -20676,9 +20015,8 @@ "displayName": "Llama 3.2 1B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 60000, - "maxOutputTokens": 60000, + "maxOutputTokens": 54000, "knowledgeCutoff": "2023-12-31", "structuredOutput": false, "lastUpdated": "2024-09-25", @@ -20696,9 +20034,8 @@ "displayName": "Llama 3.2 3B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2023-12-31", "structuredOutput": true, "lastUpdated": "2024-09-25", @@ -20716,9 +20053,8 @@ "displayName": "Llama-3.3-70B-Instruct", "description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 16384, + "maxOutputTokens": 115200, "knowledgeCutoff": "2023-12", "structuredOutput": true, "lastUpdated": "2024-12-06", @@ -20736,9 +20072,8 @@ "displayName": "Llama 4 Maverick", "description": "Open multimodal Llama model for strong reasoning and fast responses", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, - "maxOutputTokens": 16384, + "maxOutputTokens": 115200, "knowledgeCutoff": "2024-08-31", "structuredOutput": true, "lastUpdated": "2025-04-05", @@ -20756,7 +20091,6 @@ "displayName": "Llama 4 Scout", "description": "Open multimodal Llama model for long-context analysis and efficient agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1310720, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-08-31", @@ -20776,8 +20110,7 @@ "displayName": "Llama Guard 4 12B", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 1048576, + "contextWindow": 163840, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-08-31", "structuredOutput": false, @@ -20796,9 +20129,8 @@ "displayName": "Muse Glimmer 30B", "description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 16384, "knowledgeCutoff": "2026-01-04", "structuredOutput": true, "lastUpdated": "2026-08-10", @@ -20819,9 +20151,8 @@ "displayName": "Muse Spark 1.1", "description": "Open Llama multimodal model for image understanding and text reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, - "maxOutputTokens": 1048576, + "maxOutputTokens": 943718, "structuredOutput": true, "lastUpdated": "2026-07-09", "capabilities": { @@ -20833,7 +20164,7 @@ "efforts": ["minimal", "low", "medium", "high", "xhigh"] }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "video", "pdf", "audio"], "output": ["text"] } }, @@ -20841,9 +20172,8 @@ "displayName": "Muse Spark 1.2", "description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, - "maxOutputTokens": 1048576, + "maxOutputTokens": 943718, "structuredOutput": true, "lastUpdated": "2026-08-05", "capabilities": { @@ -20855,7 +20185,28 @@ "efforts": ["minimal", "low", "medium", "high", "xhigh"] }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "video", "pdf", "audio"], + "output": ["text"] + } + }, + "meta/muse-spark-1.2-contributor": { + "displayName": "Muse Spark 1.2 Contributor", + "description": "Open Llama multimodal model for image understanding and text reasoning", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 943718, + "structuredOutput": true, + "lastUpdated": "2026-08-21", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["minimal", "low", "medium", "high", "xhigh"] + }, + "modalities": { + "input": ["text", "image", "video", "pdf", "audio"], "output": ["text"] } }, @@ -20863,9 +20214,8 @@ "displayName": "Phi 4", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 16384, - "maxOutputTokens": 16384, + "maxOutputTokens": 14745, "knowledgeCutoff": "2024-06-30", "structuredOutput": true, "lastUpdated": "2025-01-10", @@ -20883,7 +20233,6 @@ "displayName": "WizardLM-2 8x22B", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65535, "maxOutputTokens": 8000, "knowledgeCutoff": "2024-04-30", @@ -20903,9 +20252,8 @@ "displayName": "MiniMax-01", "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000192, - "maxOutputTokens": 1000192, + "maxOutputTokens": 900172, "knowledgeCutoff": "2024-03-31", "structuredOutput": false, "lastUpdated": "2025-01-15", @@ -20923,7 +20271,6 @@ "displayName": "MiniMax M1", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 40000, "knowledgeCutoff": "2024-06-30", @@ -20934,6 +20281,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -20943,7 +20293,6 @@ "displayName": "MiniMax-M2", "description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, "maxOutputTokens": 131072, "structuredOutput": true, @@ -20962,7 +20311,6 @@ "displayName": "MiniMax-M2 Her", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, "maxOutputTokens": 2048, "structuredOutput": false, @@ -20981,7 +20329,6 @@ "displayName": "MiniMax-M2.1", "description": "Earlier MiniMax agent model for practical coding and productivity tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, "maxOutputTokens": 131072, "structuredOutput": false, @@ -21000,9 +20347,8 @@ "displayName": "MiniMax-M2.5", "description": "Prior MiniMax coding model for agent workflows, office edits, and automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, - "maxOutputTokens": 32768, + "maxOutputTokens": 128000, "structuredOutput": true, "lastUpdated": "2026-02-12", "capabilities": { @@ -21019,7 +20365,6 @@ "displayName": "MiniMax-M2.7", "description": "Open MiniMax flagship for coding agents, office automation, and complex environments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, "maxOutputTokens": 131072, "structuredOutput": true, @@ -21034,11 +20379,29 @@ "output": ["text"] } }, + "minimax/minimax-m2.7:free": { + "displayName": "MiniMax M2.7 (free)", + "description": "MiniMax model for chat, coding, office work, and agentic tasks", + "lifecycle": "active", + "contextWindow": 196608, + "maxOutputTokens": 176947, + "structuredOutput": false, + "lastUpdated": "2026-03-18", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "minimax/minimax-m3": { "displayName": "MiniMax-M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 512000, "structuredOutput": true, @@ -21048,8 +20411,33 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "minimax/minimax-m3:free": { + "displayName": "MiniMax M3 (free)", + "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 943718, + "structuredOutput": false, + "lastUpdated": "2026-06-01", + "isFree": true, + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -21057,9 +20445,8 @@ "displayName": "Codestral 2508", "description": "Mistral coding model for code completion, generation, and developer workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, - "maxOutputTokens": 256000, + "maxOutputTokens": 204800, "knowledgeCutoff": "2025-03-31", "structuredOutput": true, "lastUpdated": "2025-08-01", @@ -21073,32 +20460,31 @@ "output": ["text"] } }, - "mistralai/ministral-14b-2512": { - "displayName": "Ministral 3 14B 2512", - "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", + "mistralai/devstral-2512": { + "displayName": "Devstral 2", + "description": "Mistral coding agent model for repository tasks and software engineering workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 209715, + "knowledgeCutoff": "2025-12", "structuredOutput": true, - "lastUpdated": "2025-12-02", + "lastUpdated": "2025-12-09", "capabilities": { - "vision": true, + "vision": false, "reasoning": false, "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "pdf"], "output": ["text"] } }, - "mistralai/ministral-3b-2512": { - "displayName": "Ministral 3 3B 2512", + "mistralai/ministral-14b-2512": { + "displayName": "Ministral 3 14B 2512", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 131072, - "maxOutputTokens": 131072, + "contextWindow": 262144, + "maxOutputTokens": 209715, "structuredOutput": true, "lastUpdated": "2025-12-02", "capabilities": { @@ -21111,23 +20497,21 @@ "output": ["text"] } }, - "mistralai/ministral-8b": { - "displayName": "Ministral 8B", + "mistralai/ministral-3b-2512": { + "displayName": "Ministral 3 3B 2512", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 128000, - "maxOutputTokens": 128000, - "knowledgeCutoff": "2024-09-30", + "contextWindow": 131072, + "maxOutputTokens": 104857, "structuredOutput": true, - "lastUpdated": "2024-10-17", + "lastUpdated": "2025-12-02", "capabilities": { - "vision": false, + "vision": true, "reasoning": false, - "functionCalling": false + "functionCalling": true }, "modalities": { - "input": ["text"], + "input": ["text", "image"], "output": ["text"] } }, @@ -21135,9 +20519,8 @@ "displayName": "Ministral 3 8B 2512", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 209715, "structuredOutput": true, "lastUpdated": "2025-12-02", "capabilities": { @@ -21154,9 +20537,8 @@ "displayName": "Mistral Large", "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, - "maxOutputTokens": 128000, + "maxOutputTokens": 102400, "knowledgeCutoff": "2024-11-30", "structuredOutput": true, "lastUpdated": "2024-02-26", @@ -21174,9 +20556,8 @@ "displayName": "Mistral Large 2407", "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 104857, "knowledgeCutoff": "2024-03-31", "structuredOutput": true, "lastUpdated": "2024-11-19", @@ -21194,9 +20575,8 @@ "displayName": "Mistral Large 3", "description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 209715, "knowledgeCutoff": "2024-11", "structuredOutput": true, "lastUpdated": "2025-12-02", @@ -21214,9 +20594,8 @@ "displayName": "Mistral Medium 3", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 104857, "knowledgeCutoff": "2025-03-31", "structuredOutput": true, "lastUpdated": "2025-05-07", @@ -21234,9 +20613,8 @@ "displayName": "Mistral Medium 3.5", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 209715, "structuredOutput": true, "lastUpdated": "2026-04-30", "capabilities": { @@ -21256,9 +20634,8 @@ "displayName": "Mistral Medium 3.1", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 262144, + "maxOutputTokens": 104857, "knowledgeCutoff": "2025-06-30", "structuredOutput": true, "lastUpdated": "2025-08-13", @@ -21276,7 +20653,6 @@ "displayName": "Mistral Nemo", "description": "Efficient Mistral-NVIDIA open model for multilingual chat and local deployment", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-07", @@ -21296,9 +20672,8 @@ "displayName": "Saba", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, - "maxOutputTokens": 32768, + "maxOutputTokens": 26214, "knowledgeCutoff": "2024-09-30", "structuredOutput": true, "lastUpdated": "2025-02-17", @@ -21316,7 +20691,6 @@ "displayName": "Mistral Small 3", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-10-31", @@ -21336,9 +20710,8 @@ "displayName": "Mistral Small 4", "description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 209715, "knowledgeCutoff": "2025-06", "structuredOutput": true, "lastUpdated": "2026-03-16", @@ -21359,9 +20732,8 @@ "displayName": "Mistral Small 3.1 24B", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, - "maxOutputTokens": 128000, + "maxOutputTokens": 102400, "knowledgeCutoff": "2023-10-31", "structuredOutput": false, "lastUpdated": "2025-03-17", @@ -21379,8 +20751,7 @@ "displayName": "Mistral Small 3.2 24B", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 256000, + "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-10-31", "structuredOutput": true, @@ -21399,9 +20770,8 @@ "displayName": "Mixtral 8x22B Instruct", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, - "maxOutputTokens": 65536, + "maxOutputTokens": 52428, "knowledgeCutoff": "2024-01-31", "structuredOutput": true, "lastUpdated": "2024-04-17", @@ -21419,9 +20789,8 @@ "displayName": "Voxtral Small 24B 2507", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 32000, - "maxOutputTokens": 32000, + "contextWindow": 32768, + "maxOutputTokens": 26214, "structuredOutput": true, "lastUpdated": "2025-10-30", "capabilities": { @@ -21438,7 +20807,6 @@ "displayName": "Kimi K2 0711", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 100352, "knowledgeCutoff": "2024-12-31", @@ -21458,7 +20826,6 @@ "displayName": "Kimi K2 0905", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 100352, "knowledgeCutoff": "2024-12-31", @@ -21478,7 +20845,6 @@ "displayName": "Kimi K2 Thinking", "description": "Thinking Kimi model for slower research passes, planning, and hard technical questions", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 100352, "knowledgeCutoff": "2024-08", @@ -21498,9 +20864,8 @@ "displayName": "Kimi K2.5", "description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "knowledgeCutoff": "2025-01", "structuredOutput": true, "lastUpdated": "2026-01", @@ -21509,6 +20874,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -21518,9 +20886,8 @@ "displayName": "Kimi K2.6", "description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "knowledgeCutoff": "2025-01", "structuredOutput": true, "lastUpdated": "2026-04-21", @@ -21529,6 +20896,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -21538,9 +20908,8 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "knowledgeCutoff": "2025-01", "structuredOutput": true, "lastUpdated": "2026-06-12", @@ -21558,9 +20927,8 @@ "displayName": "Kimi K3", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, - "maxOutputTokens": 1048576, + "maxOutputTokens": 943718, "structuredOutput": true, "lastUpdated": "2026-07-16", "capabilities": { @@ -21573,7 +20941,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -21581,7 +20949,6 @@ "displayName": "Morph V3 Fast", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 81920, "maxOutputTokens": 38000, "structuredOutput": false, @@ -21600,7 +20967,6 @@ "displayName": "Morph V3 Large", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -21619,9 +20985,8 @@ "displayName": "Nex-N2-Mini", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-06-24", "capabilities": { @@ -21629,6 +20994,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -21638,9 +21006,8 @@ "displayName": "Nex-N2-Pro", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": false, "lastUpdated": "2026-06-08", "capabilities": { @@ -21648,6 +21015,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -21657,7 +21027,6 @@ "displayName": "Hermes 3 405B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-12-31", @@ -21677,7 +21046,6 @@ "displayName": "Hermes 3 70B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-12-31", @@ -21697,9 +21065,8 @@ "displayName": "Hermes 4 405B", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2024-08-31", "structuredOutput": false, "lastUpdated": "2025-08-26", @@ -21720,9 +21087,8 @@ "displayName": "Hermes 4 70B", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2024-08-31", "structuredOutput": false, "lastUpdated": "2025-08-26", @@ -21743,9 +21109,8 @@ "displayName": "Nemotron 3 Nano 30B A3B", "description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 228000, "structuredOutput": true, "lastUpdated": "2025-12-15", "capabilities": { @@ -21753,25 +21118,8 @@ "reasoning": true, "functionCalling": true }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, - "nvidia/nemotron-3-nano-30b-a3b:free": { - "displayName": "Nemotron 3 Nano 30B A3B (free)", - "description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 256000, - "maxOutputTokens": 256000, - "structuredOutput": false, - "lastUpdated": "2025-12-15", - "isFree": true, - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true + "thinkingOptions": { + "toggle": true }, "modalities": { "input": ["text"], @@ -21782,7 +21130,6 @@ "displayName": "Nemotron 3 Nano Omni (free)", "description": "Open Nemotron omni model combining reasoning with text, vision, and audio", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, "maxOutputTokens": 65536, "structuredOutput": false, @@ -21793,8 +21140,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "video", "audio"], "output": ["text"] } }, @@ -21802,7 +21152,6 @@ "displayName": "Nemotron 3 Super 120B A12B", "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 16384, "structuredOutput": true, @@ -21825,9 +21174,8 @@ "displayName": "Nemotron 3 Super (free)", "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-03-11", "isFree": true, @@ -21849,8 +21197,7 @@ "displayName": "Nemotron 3 Ultra 550B A55B", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 512288, + "contextWindow": 262144, "maxOutputTokens": 16384, "structuredOutput": true, "lastUpdated": "2026-06-04", @@ -21872,7 +21219,6 @@ "displayName": "Nemotron 3 Ultra (free)", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": false, @@ -21896,7 +21242,6 @@ "displayName": "Nemotron 3.5 Content Safety (free)", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 8192, "structuredOutput": false, @@ -21907,6 +21252,9 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -21916,7 +21264,6 @@ "displayName": "Nemotron 3.5 Lightning 30B A3B", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -21938,7 +21285,6 @@ "displayName": "Nemotron 3.5 Lightning (free)", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": false, @@ -21957,51 +21303,10 @@ "output": ["text"] } }, - "nvidia/nemotron-nano-12b-v2-vl:free": { - "displayName": "Nemotron Nano 12B 2 VL (free)", - "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 128000, - "maxOutputTokens": 128000, - "structuredOutput": false, - "lastUpdated": "2025-10-28", - "isFree": true, - "capabilities": { - "vision": true, - "reasoning": true, - "functionCalling": true - }, - "modalities": { - "input": ["text", "image"], - "output": ["text"] - } - }, - "nvidia/nemotron-nano-9b-v2:free": { - "displayName": "Nemotron Nano 9B V2 (free)", - "description": "Compact Nemotron model for efficient reasoning and deployable AI agents", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 128000, - "maxOutputTokens": 128000, - "structuredOutput": true, - "lastUpdated": "2025-08-18", - "isFree": true, - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "openai/gpt-3.5-turbo": { "displayName": "GPT-3.5-turbo", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 16385, "maxOutputTokens": 4096, "knowledgeCutoff": "2021-09-01", @@ -22021,9 +21326,8 @@ "displayName": "GPT-3.5 Turbo (older v0613)", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 4095, - "maxOutputTokens": 4096, + "maxOutputTokens": 3685, "knowledgeCutoff": "2021-09-30", "structuredOutput": true, "lastUpdated": "2024-01-25", @@ -22041,7 +21345,6 @@ "displayName": "GPT-3.5 Turbo 16k", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 16385, "maxOutputTokens": 4096, "knowledgeCutoff": "2021-09-30", @@ -22061,9 +21364,8 @@ "displayName": "GPT-3.5 Turbo Instruct", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 4095, - "maxOutputTokens": 4096, + "maxOutputTokens": 3685, "knowledgeCutoff": "2021-09-30", "structuredOutput": true, "lastUpdated": "2023-09-28", @@ -22081,7 +21383,6 @@ "displayName": "GPT-4", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8191, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-11", @@ -22101,7 +21402,6 @@ "displayName": "GPT-4 Turbo", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -22121,7 +21421,6 @@ "displayName": "GPT-4 Turbo Preview", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12-31", @@ -22141,7 +21440,6 @@ "displayName": "GPT-4.1", "description": "Long-lived GPT workhorse for coding, instruction following, and production apps", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -22161,7 +21459,6 @@ "displayName": "GPT-4.1 mini", "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -22181,7 +21478,6 @@ "displayName": "GPT-4.1 nano", "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -22201,7 +21497,6 @@ "displayName": "GPT-4o", "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -22221,7 +21516,6 @@ "displayName": "GPT-4o (2024-05-13)", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-09", @@ -22241,7 +21535,6 @@ "displayName": "GPT-4o (2024-08-06)", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -22261,7 +21554,6 @@ "displayName": "GPT-4o (2024-11-20)", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -22281,7 +21573,6 @@ "displayName": "GPT-4o mini", "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -22301,7 +21592,6 @@ "displayName": "GPT-4o-mini (2024-07-18)", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-10-31", @@ -22321,7 +21611,6 @@ "displayName": "GPT-5", "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22345,7 +21634,6 @@ "displayName": "GPT-5 Image", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-10-01", @@ -22365,7 +21653,6 @@ "displayName": "GPT-5 Image Mini", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -22384,7 +21671,6 @@ "displayName": "GPT-5 Mini", "description": "Small GPT-5 for responsive agents, coding help, and everyday automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22408,7 +21694,6 @@ "displayName": "GPT-5 Nano", "description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22432,7 +21717,6 @@ "displayName": "GPT-5 Pro", "description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22456,7 +21740,6 @@ "displayName": "GPT-5.1", "description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22480,7 +21763,6 @@ "displayName": "GPT-5.1 Codex", "description": "Codex GPT for repository edits, code review, and practical software agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22504,7 +21786,6 @@ "displayName": "GPT-5.1 Codex Max", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22528,7 +21809,6 @@ "displayName": "GPT-5.1 Codex mini", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22553,7 +21833,6 @@ "displayName": "GPT-5.2", "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22577,7 +21856,6 @@ "displayName": "GPT-5.2 Chat", "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-08-31", @@ -22597,7 +21875,6 @@ "displayName": "GPT-5.2 Codex", "description": "Code-specialist GPT for repository edits, reviews, and long-running software agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22621,7 +21898,6 @@ "displayName": "GPT-5.2 Pro", "description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22645,7 +21921,6 @@ "displayName": "GPT-5.3 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22669,7 +21944,6 @@ "displayName": "GPT-5.4", "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22693,7 +21967,6 @@ "displayName": "GPT-5.4 Image 2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 272000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -22715,7 +21988,6 @@ "displayName": "GPT-5.4 mini", "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22739,7 +22011,6 @@ "displayName": "GPT-5.4 nano", "description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -22763,7 +22034,6 @@ "displayName": "GPT-5.4 Pro", "description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22787,7 +22057,6 @@ "displayName": "GPT-5.5", "description": "Default frontier GPT for coding, computer use, research, and knowledge work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22811,7 +22080,6 @@ "displayName": "GPT-5.5 Pro", "description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22835,7 +22103,6 @@ "displayName": "GPT-5.6 Luna", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22859,7 +22126,6 @@ "displayName": "GPT-5.6 Luna Pro", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22883,7 +22149,6 @@ "displayName": "GPT-5.6 Sol", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22907,7 +22172,6 @@ "displayName": "GPT-5.6 Sol Pro", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22931,7 +22195,6 @@ "displayName": "GPT-5.6 Terra", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22955,7 +22218,6 @@ "displayName": "GPT-5.6 Terra Pro", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -22979,7 +22241,6 @@ "displayName": "GPT Audio", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 16384, "structuredOutput": true, @@ -22998,7 +22259,6 @@ "displayName": "GPT Audio Mini", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 16384, "structuredOutput": true, @@ -23017,7 +22277,6 @@ "displayName": "GPT Chat Latest", "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 400000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -23036,9 +22295,8 @@ "displayName": "GPT OSS 120B", "description": "Open GPT reasoning model for self-hosted agents and controllable deployments", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "structuredOutput": true, "lastUpdated": "2025-08-05", "capabilities": { @@ -23058,9 +22316,8 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "structuredOutput": true, "lastUpdated": "2025-08-05", "capabilities": { @@ -23076,34 +22333,10 @@ "output": ["text"] } }, - "openai/gpt-oss-20b:free": { - "displayName": "gpt-oss-20b (free)", - "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 131072, - "maxOutputTokens": 32768, - "structuredOutput": true, - "lastUpdated": "2025-08-05", - "isFree": true, - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "efforts": ["low", "medium", "high"] - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "openai/gpt-oss-safeguard-20b": { "displayName": "gpt-oss-safeguard-20b", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 65536, "structuredOutput": true, @@ -23125,7 +22358,6 @@ "displayName": "o1", "description": "O-series reasoning model for hard analysis, math, coding, and planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2023-09", @@ -23137,7 +22369,7 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"] + "toggle": true }, "modalities": { "input": ["text", "image", "pdf"], @@ -23148,7 +22380,6 @@ "displayName": "o1-pro", "description": "O-series reasoning model for hard analysis, math, coding, and planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2023-09", @@ -23159,6 +22390,9 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] @@ -23168,7 +22402,6 @@ "displayName": "o3", "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -23180,7 +22413,7 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"] + "toggle": true }, "modalities": { "input": ["text", "image", "pdf"], @@ -23191,7 +22424,6 @@ "displayName": "o3-mini", "description": "Smaller o-series reasoner for economical coding, math, and planning tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -23203,7 +22435,7 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"] + "toggle": true }, "modalities": { "input": ["text", "pdf"], @@ -23214,7 +22446,6 @@ "displayName": "o3 Mini High", "description": "O-series reasoning model for hard analysis, math, coding, and planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2023-10-31", @@ -23237,7 +22468,6 @@ "displayName": "o3-pro", "description": "High-effort o3 tier for difficult technical reasoning and careful answers", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -23249,7 +22479,7 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"] + "toggle": true }, "modalities": { "input": ["text", "pdf", "image"], @@ -23260,7 +22490,6 @@ "displayName": "o4-mini", "description": "Fast o-series model for compact reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -23272,7 +22501,7 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"] + "toggle": true }, "modalities": { "input": ["image", "text", "pdf"], @@ -23283,7 +22512,6 @@ "displayName": "o4 Mini High", "description": "O-series reasoning model for hard analysis, math, coding, and planning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-06-30", @@ -23306,7 +22534,6 @@ "displayName": "Auto Router", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 2000000, "maxOutputTokens": 2000000, "structuredOutput": true, @@ -23317,7 +22544,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "pdf", "video"], "output": ["text", "image"] } }, @@ -23325,7 +22552,6 @@ "displayName": "Body Builder (beta)", "description": "Preview model for early access evaluation, prototyping, and compatibility testing", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, "maxOutputTokens": 128000, "structuredOutput": false, @@ -23344,7 +22570,6 @@ "displayName": "Free Models Router", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "inputLimit": 200000, "maxOutputTokens": 8000, @@ -23365,7 +22590,6 @@ "displayName": "Fusion", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": false, @@ -23384,7 +22608,6 @@ "displayName": "Pareto Code Router", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 2000000, "maxOutputTokens": 200000, "structuredOutput": false, @@ -23403,7 +22626,6 @@ "displayName": "Perceptron Mk1", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, "maxOutputTokens": 8192, "structuredOutput": true, @@ -23413,8 +22635,11 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -23422,9 +22647,8 @@ "displayName": "Sonar", "description": "Sonar search model for current answers, retrieval, and citation-backed chat", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 127072, - "maxOutputTokens": 127072, + "maxOutputTokens": 114364, "structuredOutput": false, "lastUpdated": "2025-01-27", "capabilities": { @@ -23441,9 +22665,8 @@ "displayName": "Sonar Deep Research", "description": "Sonar search model for current answers, retrieval, and citation-backed chat", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, - "maxOutputTokens": 128000, + "maxOutputTokens": 115200, "structuredOutput": false, "lastUpdated": "2025-03-07", "capabilities": { @@ -23451,6 +22674,9 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -23460,7 +22686,6 @@ "displayName": "Sonar Pro", "description": "Advanced Sonar search model for deeper research and cited synthesis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 8000, "structuredOutput": false, @@ -23479,7 +22704,6 @@ "displayName": "Sonar Pro Search", "description": "Advanced Sonar search model for deeper research and cited synthesis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 200000, "maxOutputTokens": 8000, "structuredOutput": true, @@ -23498,9 +22722,8 @@ "displayName": "Sonar Reasoning Pro", "description": "Web-grounded reasoning model for multi-step research and cited answers", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, - "maxOutputTokens": 128000, + "maxOutputTokens": 115200, "structuredOutput": false, "lastUpdated": "2025-03-07", "capabilities": { @@ -23508,6 +22731,9 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -23517,7 +22743,6 @@ "displayName": "Laguna S 2.1", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": false, @@ -23527,6 +22752,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -23536,7 +22764,6 @@ "displayName": "Laguna S 2.1 (free)", "description": "Free provider route for experiments, demos, and cost-sensitive chat workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -23547,6 +22774,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -23556,7 +22786,6 @@ "displayName": "Laguna XS 2.1", "description": "Agentic coding model from Poolside in the XS size class for local deployment", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -23566,6 +22795,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -23575,7 +22807,6 @@ "displayName": "Laguna XS 2.1 (free)", "description": "Free provider route for experiments, demos, and cost-sensitive chat workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": false, @@ -23586,6 +22817,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -23595,7 +22829,6 @@ "displayName": "Qwen2.5 72B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, "maxOutputTokens": 16384, "knowledgeCutoff": "2024-06-30", @@ -23615,9 +22848,8 @@ "displayName": "Qwen2.5 7B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, - "maxOutputTokens": 32768, + "maxOutputTokens": 29491, "knowledgeCutoff": "2024-06-30", "structuredOutput": true, "lastUpdated": "2024-10-16", @@ -23635,9 +22867,8 @@ "displayName": "Qwen2.5 Coder 32B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, - "maxOutputTokens": 32768, + "maxOutputTokens": 29491, "knowledgeCutoff": "2024-06-30", "structuredOutput": false, "lastUpdated": "2024-11-11", @@ -23655,7 +22886,6 @@ "displayName": "Qwen Plus", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -23675,7 +22905,6 @@ "displayName": "Qwen Plus 0728", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -23691,33 +22920,12 @@ "output": ["text"] } }, - "qwen/qwen-plus-2025-07-28:thinking": { - "displayName": "Qwen Plus 0728 (thinking)", - "description": "Qwen reasoning model for deliberate problem solving, math, and coding", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 1000000, - "maxOutputTokens": 32768, - "knowledgeCutoff": "2025-03-31", - "structuredOutput": true, - "lastUpdated": "2025-09-08", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "qwen/qwen2.5-vl-72b-instruct": { "displayName": "Qwen2.5 VL 72B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 128000, - "maxOutputTokens": 128000, + "maxOutputTokens": 28800, "knowledgeCutoff": "2024-06-30", "structuredOutput": true, "lastUpdated": "2025-02-01", @@ -23735,7 +22943,6 @@ "displayName": "Qwen3 14B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-03-31", @@ -23758,7 +22965,6 @@ "displayName": "Qwen3 235B-A22B", "description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-04", @@ -23781,9 +22987,8 @@ "displayName": "Qwen3 235B A22B Instruct 2507", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 16384, + "maxOutputTokens": 235929, "knowledgeCutoff": "2025-06-30", "structuredOutput": true, "lastUpdated": "2025-07-21", @@ -23801,9 +23006,8 @@ "displayName": "Qwen3 235B A22B Thinking 2507", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 262144, - "maxOutputTokens": 32768, + "contextWindow": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2025-06-30", "structuredOutput": false, "lastUpdated": "2025-07-25", @@ -23821,9 +23025,8 @@ "displayName": "Qwen3 30B A3B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 8192, + "maxOutputTokens": 16384, "structuredOutput": false, "lastUpdated": "2025-04-28", "capabilities": { @@ -23843,7 +23046,6 @@ "displayName": "Qwen3 30B A3B Instruct 2507", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-06-30", @@ -23863,7 +23065,6 @@ "displayName": "Qwen3 30B A3B Thinking 2507", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 81920, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-06-30", @@ -23883,7 +23084,6 @@ "displayName": "Qwen3 32B", "description": "Dense open Qwen model for self-hosted chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -23906,7 +23106,6 @@ "displayName": "Qwen3 8B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-03-31", @@ -23929,7 +23128,6 @@ "displayName": "Qwen3 Coder 480B A35B", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-06-30", @@ -23949,9 +23147,8 @@ "displayName": "Qwen3-Coder 30B-A3B Instruct", "description": "Smaller Qwen coder for efficient local agents and repo-level fixes", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "knowledgeCutoff": "2025-04", "structuredOutput": true, "lastUpdated": "2025-04", @@ -23969,7 +23166,6 @@ "displayName": "Qwen3 Coder Flash", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -23989,9 +23185,8 @@ "displayName": "Qwen3 Coder Next", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "knowledgeCutoff": "2025-09", "structuredOutput": true, "lastUpdated": "2026-02-03", @@ -24009,7 +23204,6 @@ "displayName": "Qwen3 Coder Plus", "description": "Hosted Qwen coder for software agents, repo edits, and long-context code", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -24029,7 +23223,6 @@ "displayName": "Qwen3 Max", "description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -24049,7 +23242,6 @@ "displayName": "Qwen3 Max Thinking", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -24059,6 +23251,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -24068,9 +23263,8 @@ "displayName": "Qwen3-Next 80B-A3B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 16384, + "maxOutputTokens": 235929, "knowledgeCutoff": "2025-04", "structuredOutput": true, "lastUpdated": "2025-09", @@ -24088,7 +23282,6 @@ "displayName": "Qwen3-Next 80B-A3B (Thinking)", "description": "Efficient Qwen thinking model for local reasoning, math, and coding agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -24108,7 +23301,6 @@ "displayName": "Qwen3 VL 235B A22B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -24128,7 +23320,6 @@ "displayName": "Qwen3 VL 235B A22B Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -24148,9 +23339,8 @@ "displayName": "Qwen3 VL 30B A3B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 32768, + "maxOutputTokens": 16384, "knowledgeCutoff": "2025-03-31", "structuredOutput": true, "lastUpdated": "2025-10-06", @@ -24168,7 +23358,6 @@ "displayName": "Qwen3 VL 30B A3B Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-03-31", @@ -24188,7 +23377,6 @@ "displayName": "Qwen3 VL 32B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": true, @@ -24207,7 +23395,6 @@ "displayName": "Qwen3 VL 8B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 32768, "structuredOutput": true, @@ -24226,7 +23413,6 @@ "displayName": "Qwen3 VL 8B Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "structuredOutput": true, @@ -24245,9 +23431,8 @@ "displayName": "Qwen3.5 122B-A10B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 65536, + "maxOutputTokens": 81920, "structuredOutput": true, "lastUpdated": "2026-02-23", "capabilities": { @@ -24259,7 +23444,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24267,7 +23452,6 @@ "displayName": "Qwen3.5 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -24281,7 +23465,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24289,9 +23473,8 @@ "displayName": "Qwen3.5 35B-A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-02-23", "capabilities": { @@ -24303,7 +23486,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24311,7 +23494,6 @@ "displayName": "Qwen3.5 397B-A17B", "description": "Large open Qwen multimodal MoE for visual agents and long technical tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -24325,7 +23507,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24333,9 +23515,8 @@ "displayName": "Qwen3.5 9B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-02-23", "capabilities": { @@ -24347,7 +23528,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24355,7 +23536,6 @@ "displayName": "Qwen3.5-Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -24369,7 +23549,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24377,7 +23557,6 @@ "displayName": "Qwen3.5 Plus 2026-02-15", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -24392,7 +23571,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24400,7 +23579,6 @@ "displayName": "Qwen3.5 Plus 2026-04-20", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -24414,7 +23592,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24422,9 +23600,8 @@ "displayName": "Qwen3.6 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-04-22", "capabilities": { @@ -24436,7 +23613,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24444,9 +23621,8 @@ "displayName": "Qwen3.6 35B-A3B", "description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": true, "lastUpdated": "2026-04-17", "capabilities": { @@ -24458,7 +23634,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24466,7 +23642,6 @@ "displayName": "Qwen3.6 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "structuredOutput": true, @@ -24480,7 +23655,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24488,7 +23663,6 @@ "displayName": "Qwen3.6 Max Preview", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -24511,7 +23685,6 @@ "displayName": "Qwen3.6 Plus", "description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -24526,7 +23699,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24534,7 +23707,6 @@ "displayName": "Qwen3.7 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "inputLimit": 991000, "maxOutputTokens": 65536, @@ -24545,8 +23717,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24554,7 +23729,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -24576,7 +23750,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -24599,7 +23772,6 @@ "displayName": "Qwen3.8 2.4T A95B", "description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 262144, "structuredOutput": true, @@ -24621,7 +23793,6 @@ "displayName": "Qwen3.8 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -24636,7 +23807,28 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], + "output": ["text"] + } + }, + "qwen/qwen3.8-flash": { + "displayName": "Qwen3.8 Flash", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24644,7 +23836,6 @@ "displayName": "Qwen3.8 Max", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -24658,7 +23849,7 @@ "efforts": ["minimal", "low", "medium", "high", "xhigh"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24666,9 +23857,8 @@ "displayName": "Reka Edge", "description": "Multimodal model for analyzing text, images, documents, and rich media", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 16384, - "maxOutputTokens": 16384, + "maxOutputTokens": 14745, "structuredOutput": true, "lastUpdated": "2026-03-20", "capabilities": { @@ -24677,7 +23867,7 @@ "functionCalling": true }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } }, @@ -24685,9 +23875,8 @@ "displayName": "Reka Flash 3", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, - "maxOutputTokens": 65536, + "maxOutputTokens": 58982, "knowledgeCutoff": "2025-01-31", "structuredOutput": true, "lastUpdated": "2025-03-12", @@ -24705,7 +23894,6 @@ "displayName": "Relace Apply 3", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, "maxOutputTokens": 128000, "structuredOutput": false, @@ -24724,7 +23912,6 @@ "displayName": "Relace Search", "description": "Tool-capable chat model for instruction following and agentic application workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, "maxOutputTokens": 128000, "structuredOutput": false, @@ -24743,7 +23930,6 @@ "displayName": "Fugu Ultra", "description": "Quality-first multi-agent model for hard research, analysis, and competitions", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -24765,7 +23951,6 @@ "displayName": "Sakana Namazu", "description": "Multi-agent model for routing expert agents across complex analytical tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -24787,9 +23972,8 @@ "displayName": "Llama 3 8B Lunaris", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8192, - "maxOutputTokens": 16384, + "maxOutputTokens": 7372, "knowledgeCutoff": "2023-12-31", "structuredOutput": true, "lastUpdated": "2024-08-13", @@ -24807,7 +23991,6 @@ "displayName": "Llama 3.1 Euryale 70B v2.2", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-12-31", @@ -24827,7 +24010,6 @@ "displayName": "Llama 3.3 Euryale 70B", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-12-31", @@ -24843,34 +24025,10 @@ "output": ["text"] } }, - "stealth/ox-alpha": { - "displayName": "Ox Alpha", - "description": "Multimodal reasoning model for visual analysis, planning, and tool use", - "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 1048576, - "maxOutputTokens": 131072, - "structuredOutput": false, - "lastUpdated": "2026-08-20", - "isFree": true, - "capabilities": { - "vision": true, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "efforts": ["low", "high", "max"] - }, - "modalities": { - "input": ["text", "image"], - "output": ["text"] - } - }, "stepfun/step-3.5-flash": { "displayName": "Step 3.5 Flash", "description": "StepFun flash lane for quick multimodal reasoning and coding assistance", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -24890,10 +24048,9 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "inputLimit": 256000, - "maxOutputTokens": 256000, + "maxOutputTokens": 230400, "knowledgeCutoff": "2026-03-01", "structuredOutput": true, "lastUpdated": "2026-05-29", @@ -24906,7 +24063,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -24914,9 +24071,8 @@ "displayName": "Hunyuan A13B Instruct", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2025-03-31", "structuredOutput": true, "lastUpdated": "2025-07-08", @@ -24925,6 +24081,9 @@ "reasoning": true, "functionCalling": false }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -24934,7 +24093,6 @@ "displayName": "Hy-MT2-1.8B", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8192, "maxOutputTokens": 4096, "structuredOutput": false, @@ -24953,7 +24111,6 @@ "displayName": "Hy-MT2-30B-A3B", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 8192, "maxOutputTokens": 4096, "structuredOutput": true, @@ -24968,11 +24125,28 @@ "output": ["text"] } }, + "tencent/hy-mt2-7b": { + "displayName": "Hy-MT2-7B", + "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", + "lifecycle": "active", + "contextWindow": 8192, + "maxOutputTokens": 4096, + "structuredOutput": true, + "lastUpdated": "2026-08-19", + "capabilities": { + "vision": false, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "tencent/hy3": { "displayName": "Hy3", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, "maxOutputTokens": 128000, "structuredOutput": true, @@ -24994,9 +24168,8 @@ "displayName": "Hy3 preview", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 262144, - "maxOutputTokens": 262144, + "maxOutputTokens": 235929, "structuredOutput": false, "lastUpdated": "2026-04-20", "capabilities": { @@ -25012,36 +24185,36 @@ "output": ["text"] } }, - "thedrummer/cydonia-24b-v4.1": { - "displayName": "Cydonia 24B V4.1", - "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", + "tencent/hy4-preview": { + "displayName": "Hy4 preview", + "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 131072, - "maxOutputTokens": 131072, - "knowledgeCutoff": "2024-04-30", + "contextWindow": 1048576, + "maxOutputTokens": 64000, "structuredOutput": true, - "lastUpdated": "2025-09-27", + "lastUpdated": "2026-08-28", "capabilities": { "vision": false, - "reasoning": false, - "functionCalling": false + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "low", "high"] }, "modalities": { "input": ["text"], "output": ["text"] } }, - "thedrummer/rocinante-12b": { - "displayName": "Rocinante 12B", + "thedrummer/cydonia-24b-v4.1": { + "displayName": "Cydonia 24B V4.1", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 65536, - "maxOutputTokens": 65536, + "contextWindow": 131072, + "maxOutputTokens": 117964, "knowledgeCutoff": "2024-04-30", "structuredOutput": true, - "lastUpdated": "2024-09-30", + "lastUpdated": "2025-09-27", "capabilities": { "vision": false, "reasoning": false, @@ -25056,9 +24229,8 @@ "displayName": "Skyfall 36B V2", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 32768, - "maxOutputTokens": 32768, + "maxOutputTokens": 29491, "knowledgeCutoff": "2024-06-30", "structuredOutput": true, "lastUpdated": "2025-03-10", @@ -25076,9 +24248,8 @@ "displayName": "UnslopNemo 12B", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1024000, - "maxOutputTokens": 1024000, + "maxOutputTokens": 26214, "knowledgeCutoff": "2024-04-30", "structuredOutput": true, "lastUpdated": "2024-11-08", @@ -25096,9 +24267,8 @@ "displayName": "Inkling", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, - "maxOutputTokens": 262144, + "maxOutputTokens": 471859, "structuredOutput": false, "lastUpdated": "2026-07-15", "capabilities": { @@ -25118,11 +24288,54 @@ "displayName": "Inkling Small", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, "maxOutputTokens": 262144, - "structuredOutput": true, + "structuredOutput": false, + "lastUpdated": "2026-07-30", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "minimal", "low", "medium", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "audio"], + "output": ["text"] + } + }, + "thinkingmachines/inkling-small:free": { + "displayName": "Inkling Small (free)", + "description": "Multimodal reasoning model for visual analysis, planning, and tool use", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 262144, + "structuredOutput": false, "lastUpdated": "2026-07-30", + "isFree": true, + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "minimal", "low", "medium", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "audio"], + "output": ["text"] + } + }, + "thinkingmachines/inkling:free": { + "displayName": "Inkling (free)", + "description": "Multimodal reasoning model for visual analysis, planning, and tool use", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 262144, + "structuredOutput": false, + "lastUpdated": "2026-07-15", + "isFree": true, "capabilities": { "vision": true, "reasoning": true, @@ -25140,9 +24353,8 @@ "displayName": "ReMM SLERP 13B", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 6144, - "maxOutputTokens": 6144, + "maxOutputTokens": 4096, "knowledgeCutoff": "2023-06-30", "structuredOutput": true, "lastUpdated": "2023-07-22", @@ -25160,9 +24372,8 @@ "displayName": "Solar Pro 3", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, - "maxOutputTokens": 131072, + "maxOutputTokens": 117964, "structuredOutput": true, "lastUpdated": "2026-01-27", "capabilities": { @@ -25170,6 +24381,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25179,7 +24393,6 @@ "displayName": "Solar Pro 4", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 524288, "maxOutputTokens": 131072, "structuredOutput": true, @@ -25189,6 +24402,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25198,7 +24414,6 @@ "displayName": "Palmyra X5", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1040000, "maxOutputTokens": 8192, "structuredOutput": false, @@ -25217,9 +24432,8 @@ "displayName": "Grok 4.20", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 2000000, - "maxOutputTokens": 2000000, + "maxOutputTokens": 1800000, "knowledgeCutoff": "2025-09-01", "structuredOutput": true, "lastUpdated": "2026-03-31", @@ -25228,6 +24442,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] @@ -25237,9 +24454,8 @@ "displayName": "Grok 4.20 Multi-Agent", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 2000000, - "maxOutputTokens": 2000000, + "maxOutputTokens": 1800000, "knowledgeCutoff": "2025-09-01", "structuredOutput": true, "lastUpdated": "2026-03-31", @@ -25260,9 +24476,8 @@ "displayName": "Grok 4.3", "description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1000000, - "maxOutputTokens": 1000000, + "maxOutputTokens": 900000, "structuredOutput": true, "lastUpdated": "2026-04-17", "capabilities": { @@ -25282,9 +24497,8 @@ "displayName": "Grok 4.5", "description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 500000, - "maxOutputTokens": 500000, + "maxOutputTokens": 450000, "structuredOutput": true, "lastUpdated": "2026-07-08", "capabilities": { @@ -25304,9 +24518,8 @@ "displayName": "Grok 4.6", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 500000, - "maxOutputTokens": 500000, + "maxOutputTokens": 450000, "knowledgeCutoff": "2026-02-01", "structuredOutput": true, "lastUpdated": "2026-08-12", @@ -25327,9 +24540,8 @@ "displayName": "Grok Build 0.1", "description": "Fast Grok coding model tuned for agentic engineering and iterative edits", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, - "maxOutputTokens": 256000, + "maxOutputTokens": 230400, "structuredOutput": true, "lastUpdated": "2026-04-16", "capabilities": { @@ -25346,7 +24558,6 @@ "displayName": "MiMo-V2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -25361,7 +24572,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -25369,7 +24580,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1050000, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -25392,7 +24602,6 @@ "displayName": "GLM-4.5", "description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-04", @@ -25415,7 +24624,6 @@ "displayName": "GLM-4.5-Air", "description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-04", @@ -25438,7 +24646,6 @@ "displayName": "GLM-4.5V", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 65536, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -25461,9 +24668,8 @@ "displayName": "GLM-4.6", "description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, - "maxOutputTokens": 131072, + "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", "structuredOutput": true, "lastUpdated": "2025-09-30", @@ -25472,6 +24678,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25481,7 +24690,6 @@ "displayName": "GLM-4.6V", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -25492,8 +24700,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -25501,7 +24712,6 @@ "displayName": "GLM-4.7", "description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -25512,6 +24722,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25521,7 +24734,6 @@ "displayName": "GLM-4.7-Flash", "description": "Budget GLM lane for fast coding help, routing, and everyday automation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 202752, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -25532,6 +24744,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25541,7 +24756,6 @@ "displayName": "GLM-5", "description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, "maxOutputTokens": 128000, "structuredOutput": true, @@ -25551,6 +24765,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25560,7 +24777,6 @@ "displayName": "GLM-5-Turbo", "description": "Faster GLM-5 lane for coding agents that need lower latency", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 202752, "maxOutputTokens": 131072, "structuredOutput": false, @@ -25570,6 +24786,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25579,7 +24798,6 @@ "displayName": "GLM-5.1", "description": "Strong GLM coding model for agentic engineering, terminals, and repository generation", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 204800, "maxOutputTokens": 128000, "structuredOutput": true, @@ -25589,6 +24807,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { "input": ["text"], "output": ["text"] @@ -25598,9 +24819,8 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 1048576, - "maxOutputTokens": 131072, + "maxOutputTokens": 262144, "structuredOutput": true, "lastUpdated": "2026-06-13", "capabilities": { @@ -25621,9 +24841,8 @@ "displayName": "GLM 5.2 (free)", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 256000, - "maxOutputTokens": 256000, + "maxOutputTokens": 230400, "structuredOutput": true, "lastUpdated": "2026-06-13", "isFree": true, @@ -25645,10 +24864,9 @@ "displayName": "GLM-5.3", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", - "contextWindow": 1048576, + "contextWindow": 1310720, "maxOutputTokens": 131072, - "structuredOutput": false, + "structuredOutput": true, "lastUpdated": "2026-08-14", "capabilities": { "vision": false, @@ -25663,11 +24881,31 @@ "output": ["text"] } }, + "z-ai/glm-5.3-flash": { + "displayName": "GLM-5.3-Flash", + "description": "GLM vision model for visual reasoning, documents, and multimodal agents", + "lifecycle": "active", + "contextWindow": 1310720, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "video"], + "output": ["text"] + } + }, "z-ai/glm-5v-turbo": { "displayName": "GLM-5V-Turbo", "description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://openrouter.ai/models", "contextWindow": 202752, "maxOutputTokens": 131072, "structuredOutput": false, @@ -25677,8 +24915,11 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "toggle": true + }, "modalities": { - "input": ["image", "text"], + "input": ["image", "text", "video"], "output": ["text"] } } @@ -25688,7 +24929,6 @@ "displayName": "baidu/ERNIE-4.5-300B-A47B", "description": "Tool-capable chat model for instruction following and agentic application workflows", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": true, @@ -25707,7 +24947,6 @@ "displayName": "ByteDance-Seed/Seed-OSS-36B-Instruct", "description": "Tool-capable chat model for instruction following and agentic application workflows", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -25726,7 +24965,6 @@ "displayName": "deepseek-ai/DeepSeek-R1", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 164000, "maxOutputTokens": 164000, "structuredOutput": true, @@ -25745,7 +24983,6 @@ "displayName": "deepseek-ai/DeepSeek-V3", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 164000, "maxOutputTokens": 164000, "structuredOutput": true, @@ -25764,7 +25001,6 @@ "displayName": "deepseek-ai/DeepSeek-V3.1", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 164000, "maxOutputTokens": 164000, "structuredOutput": true, @@ -25786,7 +25022,6 @@ "displayName": "deepseek-ai/DeepSeek-V3.1-Terminus", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 164000, "maxOutputTokens": 164000, "structuredOutput": true, @@ -25808,7 +25043,6 @@ "displayName": "deepseek-ai/DeepSeek-V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 164000, "maxOutputTokens": 164000, "structuredOutput": true, @@ -25830,7 +25064,6 @@ "displayName": "deepseek-ai/DeepSeek-V3.2-Exp", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 164000, "maxOutputTokens": 164000, "structuredOutput": true, @@ -25852,7 +25085,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -25872,7 +25104,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -25892,7 +25123,6 @@ "displayName": "Gemma 4 26B A4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -25911,7 +25141,6 @@ "displayName": "Gemma 4 31B IT", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -25930,7 +25159,6 @@ "displayName": "inclusionAI/Ling-flash-2.0", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": true, @@ -25949,7 +25177,6 @@ "displayName": "MiniMaxAI/MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 197000, "maxOutputTokens": 131000, "structuredOutput": false, @@ -25968,7 +25195,6 @@ "displayName": "moonshotai/Kimi-K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -25987,7 +25213,6 @@ "displayName": "moonshotai/Kimi-K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26006,7 +25231,6 @@ "displayName": "openai/gpt-oss-120b", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 8000, "structuredOutput": true, @@ -26025,7 +25249,6 @@ "displayName": "openai/gpt-oss-20b", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 8000, "structuredOutput": true, @@ -26044,7 +25267,6 @@ "displayName": "Qwen/Qwen2.5-72B-Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 33000, "maxOutputTokens": 4000, "structuredOutput": true, @@ -26063,7 +25285,6 @@ "displayName": "Qwen/Qwen2.5-7B-Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 33000, "maxOutputTokens": 4000, "structuredOutput": true, @@ -26082,7 +25303,6 @@ "displayName": "Qwen/Qwen3-14B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": true, @@ -26104,7 +25324,6 @@ "displayName": "Qwen/Qwen3-235B-A22B-Thinking-2507", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26123,7 +25342,6 @@ "displayName": "Qwen/Qwen3-30B-A3B-Instruct-2507", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26142,7 +25360,6 @@ "displayName": "Qwen/Qwen3-32B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": true, @@ -26164,7 +25381,6 @@ "displayName": "Qwen/Qwen3-8B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": true, @@ -26186,7 +25402,6 @@ "displayName": "Qwen/Qwen3-Coder-30B-A3B-Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26205,7 +25420,6 @@ "displayName": "Qwen/Qwen3-Coder-480B-A35B-Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26224,7 +25438,6 @@ "displayName": "Qwen/Qwen3-VL-235B-A22B-Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26243,7 +25456,6 @@ "displayName": "Qwen/Qwen3-VL-235B-A22B-Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26262,7 +25474,6 @@ "displayName": "Qwen/Qwen3-VL-30B-A3B-Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26281,7 +25492,6 @@ "displayName": "Qwen/Qwen3-VL-30B-A3B-Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26300,7 +25510,6 @@ "displayName": "Qwen/Qwen3-VL-32B-Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26319,7 +25528,6 @@ "displayName": "Qwen/Qwen3-VL-32B-Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26338,7 +25546,6 @@ "displayName": "Qwen/Qwen3-VL-8B-Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26357,7 +25564,6 @@ "displayName": "Qwen3.5 122B-A10B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -26376,7 +25582,6 @@ "displayName": "Qwen3.5 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -26395,7 +25600,6 @@ "displayName": "Qwen3.5 35B-A3B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -26414,7 +25618,6 @@ "displayName": "Qwen3.5 397B-A17B", "description": "Large open Qwen multimodal MoE for visual agents and long technical tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -26433,7 +25636,6 @@ "displayName": "Qwen/Qwen3.5-9B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -26452,7 +25654,6 @@ "displayName": "Qwen3.6 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -26471,7 +25672,6 @@ "displayName": "Qwen3.6 35B-A3B", "description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "structuredOutput": true, @@ -26490,7 +25690,6 @@ "displayName": "stepfun-ai/Step-3.5-Flash", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26509,7 +25708,6 @@ "displayName": "tencent/Hunyuan-A13B-Instruct", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": true, @@ -26531,7 +25729,6 @@ "displayName": "Hy3 preview", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 262144, "maxOutputTokens": 262144, "lastUpdated": "2026-04-20", @@ -26549,7 +25746,6 @@ "displayName": "zai-org/GLM-4.5-Air", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 131000, "maxOutputTokens": 131000, "structuredOutput": true, @@ -26568,7 +25764,6 @@ "displayName": "zai-org/GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 205000, "maxOutputTokens": 205000, "structuredOutput": true, @@ -26587,7 +25782,6 @@ "displayName": "zai-org/GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 205000, "maxOutputTokens": 205000, "structuredOutput": true, @@ -26606,7 +25800,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 1049000, "maxOutputTokens": 262000, "structuredOutput": true, @@ -26628,7 +25821,6 @@ "displayName": "zai-org/GLM-5V-Turbo", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://cloud.siliconflow.com/models", "contextWindow": 200000, "maxOutputTokens": 131072, "lastUpdated": "2026-04-01", @@ -26651,7 +25843,6 @@ "displayName": "Step 1 (32K)", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 32768, "inputLimit": 32768, "maxOutputTokens": 32768, @@ -26671,7 +25862,6 @@ "displayName": "Step 2 (16K)", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 16384, "inputLimit": 16384, "maxOutputTokens": 8192, @@ -26691,7 +25881,6 @@ "displayName": "Step 3.5 Flash", "description": "StepFun flash lane for quick multimodal reasoning and coding assistance", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -26714,7 +25903,6 @@ "displayName": "Step 3.5 Flash 2603", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -26737,7 +25925,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -26752,7 +25939,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -26760,7 +25947,6 @@ "displayName": "Step TTS 2", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-02", @@ -26778,7 +25964,6 @@ "displayName": "StepAudio 2.5 ASR", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-02", @@ -26796,7 +25981,6 @@ "displayName": "StepAudio 2.5 TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/overview/concept", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-02", @@ -26816,7 +26000,6 @@ "displayName": "Step 1 (32K)", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 32768, "inputLimit": 32768, "maxOutputTokens": 32768, @@ -26836,7 +26019,6 @@ "displayName": "Step 2 (16K)", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 16384, "inputLimit": 16384, "maxOutputTokens": 8192, @@ -26856,7 +26038,6 @@ "displayName": "Step 3.5 Flash", "description": "StepFun flash lane for quick multimodal reasoning and coding assistance", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -26879,7 +26060,6 @@ "displayName": "Step 3.5 Flash 2603", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -26902,7 +26082,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -26917,7 +26096,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -26925,7 +26104,6 @@ "displayName": "Step TTS 2", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-02", @@ -26943,7 +26121,6 @@ "displayName": "StepAudio 2.5 ASR", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-02", @@ -26961,7 +26138,6 @@ "displayName": "StepAudio 2.5 TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/overview/concept", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-02", @@ -26981,7 +26157,6 @@ "displayName": "Step 3.5 Flash", "description": "StepFun flash lane for quick multimodal reasoning and coding assistance", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/step-plan/integrations/reasoning-api", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -27004,7 +26179,6 @@ "displayName": "Step 3.5 Flash 2603", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/step-plan/integrations/reasoning-api", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -27027,7 +26201,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.ai/docs/en/step-plan/integrations/reasoning-api", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -27042,7 +26215,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } } @@ -27052,7 +26225,6 @@ "displayName": "Step 3.5 Flash", "description": "StepFun flash lane for quick multimodal reasoning and coding assistance", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/step-plan/integrations/reasoning-api", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -27075,7 +26247,6 @@ "displayName": "Step 3.5 Flash 2603", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/step-plan/integrations/reasoning-api", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -27098,7 +26269,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/step-plan/integrations/reasoning-api", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -27113,7 +26283,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -27121,7 +26291,6 @@ "displayName": "Step Router v1", "description": "StepFun routing model that dispatches requests to the appropriate Step model.", "lifecycle": "active", - "docsUrl": "https://platform.stepfun.com/docs/zh/step-plan/integrations/reasoning-api", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -27143,7 +26312,6 @@ "displayName": "Cogito v2.1 671B", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 163840, "maxOutputTokens": 163840, "lastUpdated": "2025-11-13", @@ -27164,7 +26332,6 @@ "displayName": "DeepSeek-R1", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 163839, "maxOutputTokens": 163839, "knowledgeCutoff": "2024-07", @@ -27183,7 +26350,6 @@ "displayName": "DeepSeek-V3", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-07", @@ -27202,7 +26368,6 @@ "displayName": "DeepSeek V3.1", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-08", @@ -27224,7 +26389,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -27248,7 +26412,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Flagship DeepSeek model for coding, reasoning, and agentic work", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 512000, "maxOutputTokens": 384000, "structuredOutput": true, @@ -27271,11 +26434,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 1048576, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -27294,7 +26456,6 @@ "displayName": "Rnj-1 Instruct", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 32768, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-10", @@ -27313,7 +26474,6 @@ "displayName": "Gemma 3N E4B Instruct", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 32768, "maxOutputTokens": 32768, "structuredOutput": true, @@ -27332,7 +26492,6 @@ "displayName": "Gemma 4 31B Instruct", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01", @@ -27352,7 +26511,6 @@ "displayName": "LFM2-24B-A2B", "description": "Open-weight instruction model for adaptable chat and self-hosted production workloads", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 32768, "maxOutputTokens": 32768, "lastUpdated": "2026-02-25", @@ -27370,7 +26528,6 @@ "displayName": "Llama 3.3 70B", "description": "Compact Llama instruction model for fast chat and local deployment", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2023-12", @@ -27389,7 +26546,6 @@ "displayName": "Meta Llama 3 8B Instruct Lite", "description": "Compact Llama instruction model for fast chat and local deployment", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2024-04-18", @@ -27407,7 +26563,6 @@ "displayName": "MiniMax-M2.5", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-12", @@ -27425,7 +26580,6 @@ "displayName": "MiniMax-M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 202752, "maxOutputTokens": 131072, "structuredOutput": true, @@ -27444,7 +26598,6 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal coding model for long-context reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 524288, "maxOutputTokens": 250000, "structuredOutput": true, @@ -27463,7 +26616,6 @@ "displayName": "Kimi K2.5", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2026-01", @@ -27485,7 +26637,6 @@ "displayName": "Kimi K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 131000, "knowledgeCutoff": "2025-01", @@ -27500,7 +26651,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -27508,7 +26659,6 @@ "displayName": "Kimi K2.7 Code", "description": "Kimi coding model for software agents, refactors, and repository reasoning", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -27527,7 +26677,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -27541,7 +26690,7 @@ "efforts": ["low", "high", "max"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -27549,7 +26698,6 @@ "displayName": "Nemotron 3 Ultra 550B A55B", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 512300, "maxOutputTokens": 512300, "structuredOutput": true, @@ -27571,7 +26719,6 @@ "displayName": "GPT OSS 120B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-08", @@ -27593,7 +26740,6 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 131072, "maxOutputTokens": 131072, "structuredOutput": true, @@ -27615,7 +26761,6 @@ "displayName": "Pearl AI Gemma 4 31B Instruct", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 32000, "maxOutputTokens": 32000, "lastUpdated": "2026-04-07", @@ -27633,7 +26778,6 @@ "displayName": "Qwen 2.5 7B Instruct Turbo", "description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 32768, "maxOutputTokens": 32768, "structuredOutput": true, @@ -27652,7 +26796,6 @@ "displayName": "Qwen3 235B A22B Instruct 2507 FP8", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-07", @@ -27671,7 +26814,6 @@ "displayName": "Qwen3 Coder 480B A35B Instruct", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-04", @@ -27690,7 +26832,6 @@ "displayName": "Qwen3 Coder Next FP8", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2026-02-03", @@ -27709,7 +26850,6 @@ "displayName": "Qwen3.5 397B A17B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 130000, "lastUpdated": "2026-06-15", @@ -27730,7 +26870,6 @@ "displayName": "Qwen3.5 9B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 262144, "maxOutputTokens": 65536, "structuredOutput": true, @@ -27752,7 +26891,6 @@ "displayName": "Qwen3.6 Plus", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 1000000, "maxOutputTokens": 500000, "lastUpdated": "2026-04-30", @@ -27773,7 +26911,6 @@ "displayName": "Qwen3.7 Max", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 1000000, "maxOutputTokens": 500000, "lastUpdated": "2026-07-02", @@ -27791,7 +26928,6 @@ "displayName": "Inkling", "description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 524288, "maxOutputTokens": 131072, "structuredOutput": true, @@ -27813,7 +26949,6 @@ "displayName": "GLM-5", "description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 202752, "maxOutputTokens": 131072, "structuredOutput": true, @@ -27835,7 +26970,6 @@ "displayName": "GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "deprecated", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 202752, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-11", @@ -27858,7 +26992,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.together.ai/docs/serverless-models", "contextWindow": 512000, "maxOutputTokens": 164000, "structuredOutput": true, @@ -27876,6 +27009,48 @@ "input": ["text"], "output": ["text"] } + }, + "zai-org/GLM-5.3": { + "displayName": "GLM-5.3", + "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 262144, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "zai-org/GLM-5.3-Flash": { + "displayName": "GLM-5.3-Flash", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", + "lifecycle": "active", + "contextWindow": 1048575, + "maxOutputTokens": 400000, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "video"], + "output": ["text"] + } } }, "tencent-coding-plan": { @@ -27883,7 +27058,6 @@ "displayName": "GLM-5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 202752, "maxOutputTokens": 16384, "lastUpdated": "2026-02-11", @@ -27905,7 +27079,6 @@ "displayName": "Tencent HY 2.0 Instruct", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2026-03-08", @@ -27924,7 +27097,6 @@ "displayName": "Tencent HY 2.0 Think", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2026-03-08", @@ -27943,7 +27115,6 @@ "displayName": "Hunyuan-T1", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2026-03-08", @@ -27962,7 +27133,6 @@ "displayName": "Hunyuan-TurboS", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2026-03-08", @@ -27981,7 +27151,6 @@ "displayName": "Kimi-K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -27996,7 +27165,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -28004,7 +27173,6 @@ "displayName": "MiniMax-M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 204800, "maxOutputTokens": 32768, "lastUpdated": "2026-02-12", @@ -28023,7 +27191,6 @@ "displayName": "Auto", "description": "Automatic model router for matching prompts to suitable backends and budgets", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1772/128947", "contextWindow": 131072, "maxOutputTokens": 16384, "lastUpdated": "2026-03-08", @@ -28044,7 +27211,6 @@ "displayName": "Hy3", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1823/130060", "contextWindow": 256000, "maxOutputTokens": 64000, "lastUpdated": "2026-07-06", @@ -28055,7 +27221,28 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"], + "efforts": ["none", "high"], + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "hy4-preview": { + "displayName": "Hy4 preview", + "description": "A next-generation productivity model with significantly enhanced Agent and complex task execution capabilities.", + "lifecycle": "active", + "contextWindow": 1024000, + "maxOutputTokens": 64000, + "lastUpdated": "2026-08-28", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "high"], "toggle": true }, "modalities": { @@ -28069,7 +27256,6 @@ "displayName": "Hy3", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1823/130050", "contextWindow": 256000, "maxOutputTokens": 64000, "lastUpdated": "2026-07-06", @@ -28080,7 +27266,7 @@ "functionCalling": true }, "thinkingOptions": { - "efforts": ["low", "medium", "high"], + "efforts": ["none", "high"], "toggle": true }, "modalities": { @@ -28092,7 +27278,6 @@ "displayName": "Hy3 preview", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://cloud.tencent.com/document/product/1823/130050", "contextWindow": 256000, "maxOutputTokens": 64000, "lastUpdated": "2026-04-20", @@ -28110,6 +27295,27 @@ "input": ["text"], "output": ["text"] } + }, + "hy4-preview": { + "displayName": "Hy4 preview", + "description": "A next-generation productivity model with significantly enhanced Agent and complex task execution capabilities.", + "lifecycle": "active", + "contextWindow": 1024000, + "maxOutputTokens": 64000, + "lastUpdated": "2026-08-28", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "high"], + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } } }, "vercel": { @@ -28117,7 +27323,6 @@ "displayName": "Qwen3-14B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 40960, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -28139,7 +27344,6 @@ "displayName": "Qwen3 235B A22B Instruct 2507", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -28158,7 +27362,6 @@ "displayName": "Qwen3-30B-A3B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 40960, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -28180,7 +27383,6 @@ "displayName": "Qwen 3.32B", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-04", @@ -28202,7 +27404,6 @@ "displayName": "Qwen 3.6 Max Preview", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 240000, "maxOutputTokens": 64000, "lastUpdated": "2026-04-24", @@ -28223,7 +27424,6 @@ "displayName": "Qwen3 235B A22B Thinking 2507", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -28242,7 +27442,6 @@ "displayName": "Qwen3 Coder 480B A35B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -28261,7 +27460,6 @@ "displayName": "Qwen 3 Coder 30B A3B Instruct", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-04", @@ -28280,7 +27478,6 @@ "displayName": "Qwen3 Coder Next", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2025-09", @@ -28300,7 +27497,6 @@ "displayName": "Qwen3 Coder Plus", "description": "Hosted Qwen coder for software agents, repo edits, and long-context code", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-04", @@ -28319,7 +27515,6 @@ "displayName": "Qwen3 Embedding 0.6B", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32768, "maxOutputTokens": 32768, "lastUpdated": "2025-11-14", @@ -28337,7 +27532,6 @@ "displayName": "Qwen3 Embedding 4B", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32768, "maxOutputTokens": 32768, "lastUpdated": "2025-06-05", @@ -28355,7 +27549,6 @@ "displayName": "Qwen3 Embedding 8B", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32768, "maxOutputTokens": 32768, "lastUpdated": "2025-06-05", @@ -28373,7 +27566,6 @@ "displayName": "Qwen3 Max", "description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -28392,7 +27584,6 @@ "displayName": "Qwen3 Max Preview", "description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -28411,7 +27602,6 @@ "displayName": "Qwen 3 Max Thinking", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -28430,7 +27620,6 @@ "displayName": "Qwen3 Next 80B A3B Instruct", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -28449,7 +27638,6 @@ "displayName": "Qwen3 Next 80B A3B Thinking", "description": "Efficient Qwen thinking model for local reasoning, math, and coding agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -28468,7 +27656,6 @@ "displayName": "Qwen3 VL 235B A22B Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 129024, "knowledgeCutoff": "2025-03-31", @@ -28488,7 +27675,6 @@ "displayName": "Qwen3 VL Instruct", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 129024, "knowledgeCutoff": "2025-04", @@ -28507,7 +27693,6 @@ "displayName": "Qwen3 VL Thinking", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-09", @@ -28526,7 +27711,6 @@ "displayName": "Qwen 3.5 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "structuredOutput": true, @@ -28548,7 +27732,6 @@ "displayName": "Qwen 3.5 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04", @@ -28570,7 +27753,6 @@ "displayName": "Qwen 3.6 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "structuredOutput": true, @@ -28592,7 +27774,6 @@ "displayName": "Qwen 3.6 Plus", "description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04", @@ -28614,7 +27795,6 @@ "displayName": "Qwen 3.7 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 991000, "maxOutputTokens": 64000, "structuredOutput": true, @@ -28633,7 +27813,6 @@ "displayName": "Qwen 3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 991000, "maxOutputTokens": 64000, "lastUpdated": "2026-05-21", @@ -28654,7 +27833,6 @@ "displayName": "Qwen 3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04", @@ -28676,9 +27854,8 @@ "displayName": "Qwen3.8 2.4T A95B", "description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, - "maxOutputTokens": 131072, + "maxOutputTokens": 128000, "structuredOutput": true, "lastUpdated": "2026-08-12", "capabilities": { @@ -28698,7 +27875,6 @@ "displayName": "Qwen3.8 27B", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -28716,11 +27892,52 @@ "output": ["text"] } }, + "alibaba/qwen3.8-flash": { + "displayName": "Qwen 3.8 Flash", + "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", + "lifecycle": "active", + "contextWindow": 991000, + "maxOutputTokens": 128000, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text", "image", "pdf"], + "output": ["text"] + } + }, + "alibaba/qwen3.8-flash-next": { + "displayName": "Qwen 3.8 Flash Next", + "description": "Open-weight experimental preview of the Qwen4 architecture: hybrid-attention MoE (125B total, 6B active) with vision encoder for coding, agent tasks, and image and video understanding", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 1048576, + "structuredOutput": true, + "lastUpdated": "2026-08-27", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "xhigh"] + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "alibaba/qwen3.8-max": { "displayName": "Qwen 3.8 Max", "description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "lastUpdated": "2026-07-19", @@ -28741,7 +27958,6 @@ "displayName": "Wan v2.5 Text-to-Video Preview", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-09-24", @@ -28752,14 +27968,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "alibaba/wan-v2.6-i2v": { "displayName": "Wan v2.6 Image-to-Video", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-16", @@ -28770,14 +27985,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "alibaba/wan-v2.6-i2v-flash": { "displayName": "Wan v2.6 Image-to-Video Flash", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-16", @@ -28788,14 +28002,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "alibaba/wan-v2.6-r2v": { "displayName": "Wan v2.6 Reference-to-Video", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-16", @@ -28806,14 +28019,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "alibaba/wan-v2.6-r2v-flash": { "displayName": "Wan v2.6 Reference-to-Video Flash", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-16", @@ -28824,14 +28036,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "alibaba/wan-v2.6-t2v": { "displayName": "Wan v2.6 Text-to-Video", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-16", @@ -28842,14 +28053,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "alibaba/wan-v2.7-r2v": { "displayName": "Wan v2.7 Reference-to-Video", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-04-07", @@ -28860,14 +28070,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "alibaba/wan-v2.7-t2v": { "displayName": "Wan v2.7 Text-to-Video", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-04-07", @@ -28878,14 +28087,47 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] + } + }, + "alibaba/wan-v3.0-video": { + "displayName": "Wan v3.0 Video", + "description": "Video model for prompt-guided generation, editing, and motion workflows", + "lifecycle": "active", + "contextWindow": 0, + "maxOutputTokens": 0, + "lastUpdated": "2026-08-23", + "capabilities": { + "vision": true, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["text", "image"], + "output": ["video"] + } + }, + "alibaba/wan-v3.0-video-prime": { + "displayName": "Wan v3.0 Video Prime", + "description": "Video model for prompt-guided generation, editing, and motion workflows", + "lifecycle": "active", + "contextWindow": 0, + "maxOutputTokens": 0, + "lastUpdated": "2026-08-28", + "capabilities": { + "vision": true, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["text", "image"], + "output": ["video"] } }, "amazon/nova-2-lite": { "displayName": "Nova 2 Lite", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 1000000, "knowledgeCutoff": "2024-10", @@ -28908,7 +28150,6 @@ "displayName": "Nova Lite", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 300000, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-10", @@ -28919,7 +28160,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -28927,7 +28168,6 @@ "displayName": "Nova Micro", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-10", @@ -28946,7 +28186,6 @@ "displayName": "Nova Pro", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 300000, "maxOutputTokens": 8192, "knowledgeCutoff": "2024-10", @@ -28957,7 +28196,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -28965,7 +28204,6 @@ "displayName": "Titan Text Embeddings V2", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-04", @@ -28983,7 +28221,6 @@ "displayName": "Claude Haiku 3", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-08-31", @@ -29002,7 +28239,6 @@ "displayName": "Claude Fable 5", "description": "Claude model for creative writing, analysis, and controlled agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -29025,7 +28261,6 @@ "displayName": "Claude Haiku 4.5", "description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-02-28", @@ -29047,7 +28282,6 @@ "displayName": "Claude Opus 4", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-03-31", @@ -29066,7 +28300,6 @@ "displayName": "Claude Opus 4.5", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-05", @@ -29088,7 +28321,6 @@ "displayName": "Claude Opus 4.6", "description": "High-end Claude for difficult coding, planning, and slower expert reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05-31", @@ -29111,7 +28343,6 @@ "displayName": "Claude Opus 4.7", "description": "Stronger Opus tier for advanced software work and high-stakes reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -29134,7 +28365,6 @@ "displayName": "Claude Opus 4.8", "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01", @@ -29157,7 +28387,6 @@ "displayName": "Claude Opus 4.8 (Fast)", "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01", @@ -29180,7 +28409,6 @@ "displayName": "Claude Opus 5", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-05", @@ -29202,7 +28430,6 @@ "displayName": "Claude Opus 5 (Fast)", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-05", @@ -29224,7 +28451,6 @@ "displayName": "Claude Sonnet 4", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 8192, "knowledgeCutoff": "2025-03-31", @@ -29243,7 +28469,6 @@ "displayName": "Claude Sonnet 4.5", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-07-31", @@ -29263,7 +28488,6 @@ "displayName": "Claude Sonnet 4.6", "description": "Claude workhorse for coding agents, careful analysis, and production cost control", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -29286,7 +28510,6 @@ "displayName": "Claude Sonnet 5", "description": "Everyday Claude agent model for coding, planning, browsing, and general work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -29309,7 +28532,6 @@ "displayName": "Trinity Large Thinking", "description": "Reasoning-optimized 398B MoE agent model with extended thinking for long-horizon and multi-turn tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262100, "maxOutputTokens": 80000, "lastUpdated": "2026-04-03", @@ -29323,30 +28545,10 @@ "output": ["text"] } }, - "arcee-ai/trinity-mini": { - "displayName": "Trinity Mini", - "description": "Reasoning-tuned 26B MoE model with 3B active parameters for agents, tools, and multi-step workloads", - "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 131072, - "maxOutputTokens": 131072, - "knowledgeCutoff": "2024-10", - "lastUpdated": "2025-12", - "capabilities": { - "vision": false, - "reasoning": false, - "functionCalling": false - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "bfl/flux-2-flex": { "displayName": "FLUX.2 [flex]", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -29364,7 +28566,6 @@ "displayName": "FLUX.2 [klein] 4B", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -29382,7 +28583,6 @@ "displayName": "FLUX.2 [klein] 9B", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -29400,7 +28600,6 @@ "displayName": "FLUX.2 [max]", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 67300, "maxOutputTokens": 67300, "lastUpdated": "2026-06-08", @@ -29418,7 +28617,6 @@ "displayName": "FLUX.2 [pro]", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 67300, "maxOutputTokens": 67300, "lastUpdated": "2026-06-08", @@ -29436,7 +28634,6 @@ "displayName": "Flux 3", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-08-04", @@ -29447,14 +28644,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "bfl/flux-kontext-max": { "displayName": "FLUX.1 Kontext Max", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2025-06", @@ -29472,7 +28668,6 @@ "displayName": "FLUX.1 Kontext Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2025-06", @@ -29490,7 +28685,6 @@ "displayName": "FLUX.1 Fill [pro]", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2024-10", @@ -29508,7 +28702,6 @@ "displayName": "FLUX1.1 [pro]", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2024-10", @@ -29526,7 +28719,6 @@ "displayName": "FLUX1.1 [pro] Ultra", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2024-11", @@ -29544,7 +28736,6 @@ "displayName": "Seed 1.6", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 32000, "knowledgeCutoff": "2024-10", @@ -29566,7 +28757,6 @@ "displayName": "Seed 1.8", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2024-10", @@ -29589,7 +28779,6 @@ "displayName": "Seedance 2.0", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-04-14", @@ -29600,14 +28789,13 @@ }, "modalities": { "input": ["text", "image"], - "output": [] + "output": ["video"] } }, "bytedance/seedance-2.0-fast": { "displayName": "Seedance 2.0 Fast", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-04-14", @@ -29618,14 +28806,30 @@ }, "modalities": { "input": ["text", "image"], - "output": [] + "output": ["video"] + } + }, + "bytedance/seedance-2.0-mini": { + "displayName": "Seedance 2.0 Mini", + "description": "Video model for prompt-guided generation, editing, and motion workflows", + "lifecycle": "active", + "contextWindow": 0, + "maxOutputTokens": 0, + "lastUpdated": "2026-06-22", + "capabilities": { + "vision": true, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["text", "image"], + "output": ["video"] } }, "bytedance/seedance-2.5": { "displayName": "Seedance 2.5", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-08-07", @@ -29636,14 +28840,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "bytedance/seedance-v1.0-pro": { "displayName": "Seedance v1.0 Pro", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-06-11", @@ -29654,14 +28857,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "bytedance/seedance-v1.0-pro-fast": { "displayName": "Seedance v1.0 Pro Fast", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-10-31", @@ -29672,14 +28874,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "bytedance/seedance-v1.5-pro": { "displayName": "Seedance v1.5 Pro", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-16", @@ -29690,14 +28891,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "bytedance/seedream-4.0": { "displayName": "Seedream 4.0", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-08-28", @@ -29715,7 +28915,6 @@ "displayName": "Seedream 4.5", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-11-28", @@ -29733,7 +28932,6 @@ "displayName": "Seedream 5.0 Lite", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-01-28", @@ -29751,7 +28949,6 @@ "displayName": "Seedream 5.0 Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-11", @@ -29769,7 +28966,6 @@ "displayName": "Command A", "description": "Cohere command model for multilingual enterprise agents, tools, and chat", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 8000, "knowledgeCutoff": "2024-10", @@ -29788,7 +28984,6 @@ "displayName": "Embed v4.0", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 1536, "lastUpdated": "2025-04-15", @@ -29806,7 +29001,6 @@ "displayName": "Cohere Rerank 3.5", "description": "Reranking model for improving retrieval quality in search and recommendation systems", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 4096, "maxOutputTokens": 4096, "lastUpdated": "2024-12-02", @@ -29824,7 +29018,6 @@ "displayName": "Cohere Rerank 4 Fast", "description": "Reranking model for improving retrieval quality in search and recommendation systems", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 32000, "lastUpdated": "2025-12-11", @@ -29842,7 +29035,6 @@ "displayName": "Cohere Rerank 4 Pro", "description": "Reranking model for improving retrieval quality in search and recommendation systems", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 32000, "lastUpdated": "2025-12-11", @@ -29860,7 +29052,6 @@ "displayName": "DeepSeek-R1", "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-07", @@ -29879,7 +29070,6 @@ "displayName": "DeepSeek V3 0324", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 163840, "maxOutputTokens": 163840, "lastUpdated": "2024-12-26", @@ -29897,7 +29087,6 @@ "displayName": "DeepSeek-V3.1", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 163840, "maxOutputTokens": 128000, "lastUpdated": "2025-08-21", @@ -29918,7 +29107,6 @@ "displayName": "DeepSeek V3.1 Terminus", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-07", @@ -29940,7 +29128,6 @@ "displayName": "DeepSeek V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 8000, "knowledgeCutoff": "2024-07", @@ -29960,7 +29147,6 @@ "displayName": "DeepSeek V3.2 Thinking", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 8000, "knowledgeCutoff": "2024-07", @@ -29979,7 +29165,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -30003,7 +29188,6 @@ "displayName": "DeepSeek V4 Flash 0731", "description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -30019,13 +29203,34 @@ "output": ["text"] } }, + "deepseek/deepseek-v4-flash-vision-exp": { + "displayName": "DeepSeek V4 Flash Vision Exp", + "description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 384000, + "structuredOutput": true, + "lastUpdated": "2026-08-21", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["high", "xhigh"], + "toggle": true + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "deepseek/deepseek-v4-pro": { "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 1048600, - "maxOutputTokens": 1048600, + "contextWindow": 1000000, + "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", "structuredOutput": true, "lastUpdated": "2026-04-24", @@ -30047,11 +29252,10 @@ "displayName": "DeepSeek V4 Pro 0813", "description": "Flagship DeepSeek model for coding, reasoning, and agentic work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 384000, "structuredOutput": true, - "lastUpdated": "2026-08-12", + "lastUpdated": "2026-08-22", "capabilities": { "vision": false, "reasoning": true, @@ -30070,7 +29274,6 @@ "displayName": "S1", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-10-20", @@ -30088,7 +29291,6 @@ "displayName": "S1 (Free)", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-10-20", @@ -30106,7 +29308,6 @@ "displayName": "S2 Pro", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-09", @@ -30124,7 +29325,6 @@ "displayName": "S2 Pro (Free)", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-09", @@ -30142,7 +29342,6 @@ "displayName": "S2.1 Pro", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-28", @@ -30160,7 +29359,6 @@ "displayName": "S2.1 Pro (Free)", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-28", @@ -30178,7 +29376,6 @@ "displayName": "Transcribe-1", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-01", @@ -30196,7 +29393,6 @@ "displayName": "Transcribe-1 (Free)", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-01", @@ -30214,7 +29410,6 @@ "displayName": "Gemini 2.5 Flash", "description": "Fast Gemini workhorse for multimodal apps where latency and price matter", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -30229,7 +29424,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -30237,7 +29432,6 @@ "displayName": "Nano Banana (Gemini 2.5 Flash Image)", "description": "Nano Banana image model for fast generation, edits, and character-consistent assets", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32768, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -30256,7 +29450,6 @@ "displayName": "Gemini 2.5 Flash Lite", "description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -30279,7 +29472,6 @@ "displayName": "Gemini 2.5 Pro", "description": "Google's proven reasoning model for coding, math, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -30291,7 +29483,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -30299,7 +29491,6 @@ "displayName": "Gemini 3 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 65000, "knowledgeCutoff": "2025-03", @@ -30321,7 +29512,6 @@ "displayName": "Nano Banana Pro", "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 65536, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -30340,7 +29530,6 @@ "displayName": "Nano Banana 2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -30362,7 +29551,6 @@ "displayName": "Gemini 3.1 Flash Image Preview (Nano Banana 2)", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -30384,7 +29572,6 @@ "displayName": "Gemini 3.1 Flash Lite", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 65000, "knowledgeCutoff": "2025-01", @@ -30407,7 +29594,6 @@ "displayName": "Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite)", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 65536, "maxOutputTokens": 4096, "knowledgeCutoff": "2025-01", @@ -30427,7 +29613,6 @@ "displayName": "Gemini 3.1 Pro Preview", "description": "Reasoning-first Gemini preview for agentic coding and complex problem solving", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01", @@ -30450,7 +29635,6 @@ "displayName": "Gemini 3.5 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01", @@ -30473,7 +29657,6 @@ "displayName": "Gemini 3.5 Flash Lite", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 65000, "knowledgeCutoff": "2026-03", @@ -30492,11 +29675,44 @@ "output": ["text"] } }, + "google/gemini-3.5-transcribe": { + "displayName": "Gemini 3.5 Transcribe", + "description": "Speech transcription model for accurate audio-to-text and captioning workflows", + "lifecycle": "active", + "contextWindow": 0, + "maxOutputTokens": 0, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": false, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["audio"], + "output": ["text"] + } + }, + "google/gemini-3.5-transcribe-live": { + "displayName": "Gemini 3.5 Transcribe Live", + "description": "Speech transcription model for accurate audio-to-text and captioning workflows", + "lifecycle": "active", + "contextWindow": 0, + "maxOutputTokens": 0, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": false, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["audio"], + "output": ["text"] + } + }, "google/gemini-3.6-flash": { "displayName": "Gemini 3.6 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2026-03", @@ -30519,7 +29735,6 @@ "displayName": "Gemini 3.7 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 65536, "knowledgeCutoff": "2026-03", @@ -30542,7 +29757,6 @@ "displayName": "Gemini Embedding 001", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "knowledgeCutoff": "2025-05", @@ -30561,7 +29775,6 @@ "displayName": "Gemini Embedding 2", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "knowledgeCutoff": "2025-11", @@ -30580,7 +29793,6 @@ "displayName": "Gemini Omni Flash Preview", "description": "Omni-modal model for text, vision, audio, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 57920, "lastUpdated": "2026-06-30", @@ -30598,7 +29810,6 @@ "displayName": "Gemma 4 26B A4B IT", "description": "Open Gemma instruction model for efficient chat and self-hosted deployments", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -30617,7 +29828,6 @@ "displayName": "Gemma 4 31B IT", "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -30636,7 +29846,6 @@ "displayName": "Text Embedding 005", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-08", @@ -30654,7 +29863,6 @@ "displayName": "Text Multilingual Embedding 002", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-03", @@ -30672,7 +29880,6 @@ "displayName": "Veo 3.0 Fast Generate", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -30683,14 +29890,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "google/veo-3.0-generate-001": { "displayName": "Veo 3.0", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -30701,14 +29907,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "google/veo-3.1-fast-generate-001": { "displayName": "Veo 3.1 Fast Generate", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -30719,14 +29924,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "google/veo-3.1-generate-001": { "displayName": "Veo 3.1", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -30737,14 +29941,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "google/veo-3.1-lite-generate-001": { "displayName": "Veo 3.1 Lite Generate", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-04-02", @@ -30755,14 +29958,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "inception/mercury-2": { "displayName": "Mercury 2", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 128000, "lastUpdated": "2026-03-06", @@ -30783,7 +29985,6 @@ "displayName": "Mercury Coder Small Beta", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 16384, "lastUpdated": "2025-02-26", @@ -30801,7 +30002,6 @@ "displayName": "Ling 3.0 Flash", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 32000, "lastUpdated": "2026-08-06", @@ -30815,11 +30015,52 @@ "output": ["text"] } }, + "inclusionai/ling-3.0-flash-fin": { + "displayName": "Ling 3.0 Flash Fin", + "description": "Finance-enhanced model for financial research, multi-step investment workflows, and long-horizon planning and execution", + "lifecycle": "active", + "contextWindow": 256000, + "maxOutputTokens": 32000, + "lastUpdated": "2026-08-27", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "inclusionai/ling-3.0-flash-fin-free": { + "displayName": "Ling 3.0 Flash Fin (Free)", + "description": "Finance-enhanced model for financial research, multi-step investment workflows, and long-horizon planning and execution", + "lifecycle": "active", + "contextWindow": 256000, + "maxOutputTokens": 32000, + "lastUpdated": "2026-08-27", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "interfaze/interfaze-beta": { "displayName": "Interfaze Beta", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 32000, "lastUpdated": "2026-04-29", @@ -30840,7 +30081,6 @@ "displayName": "Kling v2.5 Turbo Image-to-Video", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-09-23", @@ -30851,14 +30091,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "klingai/kling-v2.5-turbo-t2v": { "displayName": "Kling v2.5 Turbo Text-to-Video", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-09-23", @@ -30869,14 +30108,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "klingai/kling-v2.6-i2v": { "displayName": "Kling v2.6 Image-to-Video", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-21", @@ -30887,14 +30125,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "klingai/kling-v2.6-motion-control": { "displayName": "Kling v2.6 Motion Control", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-21", @@ -30905,14 +30142,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "klingai/kling-v2.6-t2v": { "displayName": "Kling v2.6 Text-to-Video", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-12-21", @@ -30923,14 +30159,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "klingai/kling-v3.0-i2v": { "displayName": "Kling v3.0 Image-to-Video", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-02-05", @@ -30941,14 +30176,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "klingai/kling-v3.0-motion-control": { "displayName": "Kling v3.0 Motion Control", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-04", @@ -30959,14 +30193,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "klingai/kling-v3.0-t2v": { "displayName": "Kling v3.0 Text-to-Video", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-02-05", @@ -30977,14 +30210,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "kwaipilot/kat-coder-air-v2.5": { "displayName": "Kat Coder Air V2.5", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 80000, "lastUpdated": "2026-07-10", @@ -31002,7 +30234,6 @@ "displayName": "KAT-Coder-Pro V1", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 32000, "knowledgeCutoff": "2024-10", @@ -31021,7 +30252,6 @@ "displayName": "Kat Coder Pro V2", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "lastUpdated": "2026-03-30", @@ -31039,7 +30269,6 @@ "displayName": "Kat Coder Pro V2.5", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 80000, "lastUpdated": "2026-07-10", @@ -31057,7 +30286,6 @@ "displayName": "Llama 3.1 70B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 8192, "knowledgeCutoff": "2023-12", @@ -31076,7 +30304,6 @@ "displayName": "Llama 3.1 8B Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 8192, "knowledgeCutoff": "2023-12", @@ -31095,7 +30322,6 @@ "displayName": "Llama-3.3-70B-Instruct", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -31115,7 +30341,6 @@ "displayName": "Llama-4-Maverick-17B-128E-Instruct-FP8", "description": "Open multimodal Llama model for strong reasoning and fast responses", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-08", @@ -31135,7 +30360,6 @@ "displayName": "Llama-4-Scout-17B-16E-Instruct-FP8", "description": "Open multimodal Llama model for long-context analysis and efficient agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-08", @@ -31155,7 +30379,6 @@ "displayName": "Muse Glimmer 30B", "description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2026-01-04", @@ -31174,11 +30397,27 @@ "output": ["text"] } }, + "meta/muse-image-1.0": { + "displayName": "Muse Image 1.0", + "description": "Image model for prompt-driven generation, editing, and visual design workflows", + "lifecycle": "active", + "contextWindow": 0, + "maxOutputTokens": 0, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["text", "image"], + "output": ["image"] + } + }, "meta/muse-spark-1.1": { "displayName": "Muse Spark 1.1", "description": "Open Llama instruction model for multilingual chat, reasoning, and coding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1048576, "maxOutputTokens": 1048576, "structuredOutput": true, @@ -31200,7 +30439,6 @@ "displayName": "Muse Spark 1.2", "description": "Open Llama multimodal model for image understanding and text reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1048576, "maxOutputTokens": 1048576, "structuredOutput": true, @@ -31219,7 +30457,6 @@ "displayName": "Muse Spark 1.2 Contributor", "description": "Open Llama multimodal model for image understanding and text reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1048576, "maxOutputTokens": 1048576, "lastUpdated": "2026-08-05", @@ -31237,7 +30474,6 @@ "displayName": "MiniMax H3", "description": "Video model for prompt-guided generation, editing, and motion workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-30", @@ -31248,14 +30484,30 @@ }, "modalities": { "input": ["text", "image"], - "output": [] + "output": ["video"] + } + }, + "minimax/minimax-h3-max": { + "displayName": "MiniMax H3 Max", + "description": "Video model for prompt-guided generation, editing, and motion workflows", + "lifecycle": "active", + "contextWindow": 0, + "maxOutputTokens": 0, + "lastUpdated": "2026-08-27", + "capabilities": { + "vision": true, + "reasoning": false, + "functionCalling": false + }, + "modalities": { + "input": ["text", "image"], + "output": ["video"] } }, "minimax/minimax-m2": { "displayName": "MiniMax M2", "description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 205000, "maxOutputTokens": 205000, "knowledgeCutoff": "2024-10", @@ -31274,7 +30526,6 @@ "displayName": "MiniMax M2.1", "description": "Earlier MiniMax agent model for practical coding and productivity tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2025-12-23", @@ -31292,7 +30543,6 @@ "displayName": "MiniMax M2.1 Lightning", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-10", @@ -31311,7 +30561,6 @@ "displayName": "MiniMax M2.5", "description": "Prior MiniMax coding model for agent workflows, office edits, and automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 204800, "maxOutputTokens": 131000, "lastUpdated": "2026-02-12", @@ -31329,7 +30578,6 @@ "displayName": "MiniMax M2.5 High Speed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 204800, "maxOutputTokens": 131000, "lastUpdated": "2026-02-13", @@ -31347,7 +30595,6 @@ "displayName": "Minimax M2.7", "description": "Open MiniMax flagship for coding agents, office automation, and complex environments", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 204800, "maxOutputTokens": 131000, "lastUpdated": "2026-03-18", @@ -31361,11 +30608,28 @@ "output": ["text"] } }, + "minimax/minimax-m2.7-free": { + "displayName": "Minimax M2.7 (Free)", + "description": "Open MiniMax flagship for coding agents, office automation, and complex environments", + "lifecycle": "active", + "contextWindow": 196608, + "maxOutputTokens": 196608, + "lastUpdated": "2026-03-18", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "minimax/minimax-m2.7-highspeed": { "displayName": "MiniMax M2.7 High Speed", "description": "Low-latency M2.7 variant for interactive coding plans and agent loops", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 204800, "maxOutputTokens": 131100, "lastUpdated": "2026-03-18", @@ -31383,9 +30647,8 @@ "displayName": "MiniMax M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 1000000, - "maxOutputTokens": 1000000, + "contextWindow": 512000, + "maxOutputTokens": 512000, "lastUpdated": "2026-06-01", "capabilities": { "vision": true, @@ -31400,11 +30663,31 @@ "output": ["text"] } }, + "minimax/minimax-m3-free": { + "displayName": "MiniMax M3 (Free)", + "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 1048576, + "lastUpdated": "2026-06-01", + "isFree": true, + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "mistral/codestral": { "displayName": "Codestral (latest)", "description": "Mistral code model for completions, refactors, and developer IDE workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 4096, "knowledgeCutoff": "2024-10", @@ -31423,7 +30706,6 @@ "displayName": "Codestral Embed", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2025-05-28", @@ -31441,7 +30723,6 @@ "displayName": "Devstral 2", "description": "Mistral coding agent model for repository tasks and software engineering workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2024-10", @@ -31460,7 +30741,6 @@ "displayName": "Devstral Small 2", "description": "Mistral coding agent model for repository tasks and software engineering workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2024-10", @@ -31475,49 +30755,10 @@ "output": ["text"] } }, - "mistral/magistral-medium": { - "displayName": "Magistral Medium (latest)", - "description": "Mistral reasoning model for transparent analysis, math, and complex decisions", - "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 128000, - "maxOutputTokens": 16384, - "knowledgeCutoff": "2025-06", - "lastUpdated": "2025-03-20", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, - "mistral/magistral-small": { - "displayName": "Magistral Small", - "description": "Mistral reasoning model for transparent analysis, math, and complex decisions", - "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 128000, - "maxOutputTokens": 128000, - "knowledgeCutoff": "2025-06", - "lastUpdated": "2025-03-17", - "capabilities": { - "vision": false, - "reasoning": true, - "functionCalling": true - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "mistral/ministral-14b": { "displayName": "Ministral 14B", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2024-10", @@ -31536,7 +30777,6 @@ "displayName": "Ministral 3B (latest)", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-10", @@ -31555,7 +30795,6 @@ "displayName": "Ministral 8B (latest)", "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-10", @@ -31574,7 +30813,6 @@ "displayName": "Mistral Embed", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2023-12-11", @@ -31592,7 +30830,6 @@ "displayName": "Mistral Large 3", "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "knowledgeCutoff": "2024-10", @@ -31611,7 +30848,6 @@ "displayName": "Mistral Medium 3.1", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2024-10", @@ -31630,7 +30866,6 @@ "displayName": "Mistral Medium Latest", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "lastUpdated": "2026-05-21", @@ -31651,7 +30886,6 @@ "displayName": "Mistral Nemo", "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-04", @@ -31670,7 +30904,6 @@ "displayName": "Mistral Small (latest)", "description": "Efficient Mistral model for fast chat, extraction, and production assistants", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 4000, "knowledgeCutoff": "2025-06", @@ -31689,7 +30922,6 @@ "displayName": "Pixtral 12B", "description": "Mistral vision-language model for image understanding and multimodal chat", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 128000, "knowledgeCutoff": "2024-09", @@ -31708,7 +30940,6 @@ "displayName": "Kimi K2 Instruct", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 131072, "lastUpdated": "2025-09-05", @@ -31726,7 +30957,6 @@ "displayName": "Kimi K2 Thinking", "description": "Thinking Kimi model for slower research passes, planning, and hard technical questions", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 216144, "maxOutputTokens": 216144, "knowledgeCutoff": "2024-08", @@ -31745,7 +30975,6 @@ "displayName": "Kimi K2.5", "description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262114, "maxOutputTokens": 262114, "knowledgeCutoff": "2025-01", @@ -31768,7 +30997,6 @@ "displayName": "Kimi K2.6", "description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262000, "maxOutputTokens": 262000, "knowledgeCutoff": "2025-01", @@ -31791,7 +31019,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -31811,7 +31038,6 @@ "displayName": "Kimi K2.7 Code High Speed", "description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-01", @@ -31831,7 +31057,6 @@ "displayName": "Kimi K3", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -31850,7 +31075,6 @@ "displayName": "Kimi K3 Fast", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -31872,7 +31096,6 @@ "displayName": "Morph v3 Fast", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 16000, "maxOutputTokens": 16000, "lastUpdated": "2024-08-15", @@ -31890,7 +31113,6 @@ "displayName": "Morph v3 Large", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 32000, "lastUpdated": "2024-08-15", @@ -31908,7 +31130,6 @@ "displayName": "Nemotron 3 Nano 30B A3B", "description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 262144, "lastUpdated": "2025-12-15", @@ -31929,7 +31150,6 @@ "displayName": "NVIDIA Nemotron 3 Super 120B A12B", "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 32000, "lastUpdated": "2026-03-11", @@ -31950,7 +31170,6 @@ "displayName": "Nemotron 3 Ultra", "description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 65000, "lastUpdated": "2026-06-04", @@ -31971,7 +31190,6 @@ "displayName": "Nemotron 3.5 Lightning 30B", "description": "Nemotron model for efficient reasoning, coding, and specialized AI agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262144, "maxOutputTokens": 131072, "structuredOutput": true, @@ -31993,7 +31211,6 @@ "displayName": "Nvidia Nemotron Nano 12B V2 VL", "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 131072, "lastUpdated": "2025-10-28", @@ -32014,7 +31231,6 @@ "displayName": "Nvidia Nemotron Nano 9B V2", "description": "Compact Nemotron model for efficient reasoning and deployable AI agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 131072, "lastUpdated": "2025-08-18", @@ -32035,7 +31251,6 @@ "displayName": "GPT-3.5 Turbo", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 16385, "inputLimit": 12289, "maxOutputTokens": 4096, @@ -32056,7 +31271,6 @@ "displayName": "GPT-4 Turbo", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "deprecated", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 4096, "knowledgeCutoff": "2023-12", @@ -32076,7 +31290,6 @@ "displayName": "GPT-4.1", "description": "Long-lived GPT workhorse for coding, instruction following, and production apps", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -32096,7 +31309,6 @@ "displayName": "GPT-4.1 (Fast)", "description": "Long-lived GPT workhorse for coding, instruction following, and production apps", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1047576, "inputLimit": 1014808, "maxOutputTokens": 32768, @@ -32117,7 +31329,6 @@ "displayName": "GPT-4.1 mini", "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -32137,7 +31348,6 @@ "displayName": "GPT-4.1 mini (Fast)", "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1047576, "inputLimit": 1014808, "maxOutputTokens": 32768, @@ -32158,7 +31368,6 @@ "displayName": "GPT-4.1 nano", "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks", "lifecycle": "deprecated", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1047576, "maxOutputTokens": 32768, "knowledgeCutoff": "2024-04", @@ -32178,7 +31387,6 @@ "displayName": "GPT-4.1 nano (Fast)", "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1047576, "inputLimit": 1014808, "maxOutputTokens": 32768, @@ -32199,7 +31407,6 @@ "displayName": "GPT-4o", "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -32219,7 +31426,6 @@ "displayName": "GPT-4o (Fast)", "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "inputLimit": 111616, "maxOutputTokens": 16384, @@ -32240,7 +31446,6 @@ "displayName": "GPT-4o mini", "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 16384, "knowledgeCutoff": "2023-09", @@ -32260,7 +31465,6 @@ "displayName": "GPT-4o mini (Fast)", "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "inputLimit": 111616, "maxOutputTokens": 16384, @@ -32277,32 +31481,10 @@ "output": ["text"] } }, - "openai/gpt-4o-mini-search-preview": { - "displayName": "GPT 4o Mini Search Preview", - "description": "Compact GPT model for low-latency assistance and high-volume workloads", - "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 128000, - "inputLimit": 111616, - "maxOutputTokens": 16384, - "knowledgeCutoff": "2023-09", - "structuredOutput": false, - "lastUpdated": "2025-01", - "capabilities": { - "vision": false, - "reasoning": false, - "functionCalling": false - }, - "modalities": { - "input": ["text"], - "output": ["text"] - } - }, "openai/gpt-4o-mini-transcribe": { "displayName": "GPT-4o mini Transcribe", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2024-03-13", @@ -32320,7 +31502,6 @@ "displayName": "GPT-4o Transcribe", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2024-03-13", @@ -32338,7 +31519,6 @@ "displayName": "GPT-5", "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32362,7 +31542,6 @@ "displayName": "GPT-5-Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32374,6 +31553,9 @@ "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "efforts": ["low", "medium", "high"] + }, "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] @@ -32383,7 +31565,6 @@ "displayName": "GPT-5 (Fast)", "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32407,7 +31588,6 @@ "displayName": "GPT-5 Mini", "description": "Small GPT-5 for responsive agents, coding help, and everyday automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32431,7 +31611,6 @@ "displayName": "GPT-5 mini (Fast)", "description": "Small GPT-5 for responsive agents, coding help, and everyday automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32455,7 +31634,6 @@ "displayName": "GPT-5 Nano", "description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32479,7 +31657,6 @@ "displayName": "GPT-5 pro", "description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 128000, "maxOutputTokens": 272000, @@ -32503,7 +31680,6 @@ "displayName": "GPT-5.1-Codex", "description": "Codex GPT for repository edits, code review, and practical software agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32527,7 +31703,6 @@ "displayName": "GPT 5.1 Codex Max", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32551,7 +31726,6 @@ "displayName": "GPT-5.1 Codex mini", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32575,7 +31749,6 @@ "displayName": "GPT 5.1 Thinking", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32598,7 +31771,6 @@ "displayName": "GPT 5.1 Thinking (Fast)", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32617,7 +31789,6 @@ "displayName": "GPT-5.2", "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32641,7 +31812,6 @@ "displayName": "GPT-5.2-Codex", "description": "Code-specialist GPT for repository edits, reviews, and long-running software agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32665,7 +31835,6 @@ "displayName": "GPT 5.2 (Fast)", "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32689,7 +31858,6 @@ "displayName": "GPT 5.2 ", "description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32713,7 +31881,6 @@ "displayName": "GPT 5.3 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32737,7 +31904,6 @@ "displayName": "GPT 5.3 Codex (Fast)", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32761,7 +31927,6 @@ "displayName": "GPT 5.4", "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -32785,7 +31950,6 @@ "displayName": "GPT 5.4 (Fast)", "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -32809,7 +31973,6 @@ "displayName": "GPT 5.4 Mini", "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32833,7 +31996,6 @@ "displayName": "GPT 5.4 Mini (Fast)", "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32857,7 +32019,6 @@ "displayName": "GPT 5.4 Nano", "description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 400000, "inputLimit": 272000, "maxOutputTokens": 128000, @@ -32881,7 +32042,6 @@ "displayName": "GPT 5.4 Pro", "description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -32905,7 +32065,6 @@ "displayName": "GPT 5.5", "description": "Default frontier GPT for coding, computer use, research, and knowledge work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "inputLimit": 872000, "maxOutputTokens": 128000, @@ -32929,7 +32088,6 @@ "displayName": "GPT 5.5 (Fast)", "description": "Default frontier GPT for coding, computer use, research, and knowledge work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "inputLimit": 872000, "maxOutputTokens": 128000, @@ -32953,7 +32111,6 @@ "displayName": "GPT 5.5 Pro", "description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "inputLimit": 872000, "maxOutputTokens": 128000, @@ -32977,7 +32134,6 @@ "displayName": "GPT 5.6 Luna", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -33001,7 +32157,6 @@ "displayName": "GPT 5.6 Luna (Fast)", "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -33025,7 +32180,6 @@ "displayName": "GPT 5.6 Sol", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -33049,7 +32203,6 @@ "displayName": "GPT 5.6 Sol (Fast)", "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -33073,7 +32226,6 @@ "displayName": "GPT 5.6 Terra", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -33097,7 +32249,6 @@ "displayName": "GPT 5.6 Terra (Fast)", "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -33121,7 +32272,6 @@ "displayName": "GPT Image 1", "description": "OpenAI image model for production generation, edits, and brand-safe visual workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-04-24", @@ -33139,7 +32289,6 @@ "displayName": "GPT Image 1 Mini", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-10-06", @@ -33157,7 +32306,6 @@ "displayName": "GPT Image 1.5", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-11-25", @@ -33175,7 +32323,6 @@ "displayName": "GPT Image 2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-04-21", @@ -33193,7 +32340,6 @@ "displayName": "GPT OSS 120B", "description": "Open GPT reasoning model for self-hosted agents and controllable deployments", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-10", @@ -33216,7 +32362,6 @@ "displayName": "GPT OSS 20B", "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "inputLimit": 122880, "maxOutputTokens": 8192, @@ -33236,14 +32381,35 @@ "output": ["text"] } }, + "openai/gpt-oss-safeguard-120b": { + "displayName": "GPT OSS Safeguard 120B", + "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", + "lifecycle": "active", + "contextWindow": 128000, + "inputLimit": 112000, + "maxOutputTokens": 16000, + "structuredOutput": true, + "lastUpdated": "2025-10-29", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "medium", "high"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "openai/gpt-oss-safeguard-20b": { "displayName": "gpt-oss-safeguard-20b", "description": "Safety model for policy screening, moderation, and risk-aware routing workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 131072, - "inputLimit": 65536, - "maxOutputTokens": 65536, + "contextWindow": 128000, + "inputLimit": 112000, + "maxOutputTokens": 16000, "knowledgeCutoff": "2024-10", "lastUpdated": "2024-12-01", "capabilities": { @@ -33263,7 +32429,6 @@ "displayName": "GPT-Realtime-1.5", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-02-23", @@ -33281,7 +32446,6 @@ "displayName": "gpt-realtime-2", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-05-07", @@ -33299,7 +32463,6 @@ "displayName": "gpt-realtime-2.1", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "inputLimit": 96000, "maxOutputTokens": 32000, @@ -33323,7 +32486,6 @@ "displayName": "GPT-Realtime mini", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2025-10-10", @@ -33341,7 +32503,6 @@ "displayName": "gpt-realtime-whisper", "description": "Streaming speech-to-text model for low-latency transcript deltas from live audio", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-05-07", @@ -33359,7 +32520,6 @@ "displayName": "o1", "description": "O-series reasoning model for hard analysis, math, coding, and planning", "lifecycle": "deprecated", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2023-09", @@ -33382,7 +32542,6 @@ "displayName": "o3", "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -33401,34 +32560,10 @@ "output": ["text"] } }, - "openai/o3-deep-research": { - "displayName": "o3-deep-research", - "description": "Research model for long-horizon investigation, synthesis, and analytical reports", - "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 200000, - "inputLimit": 100000, - "maxOutputTokens": 100000, - "knowledgeCutoff": "2024-05", - "lastUpdated": "2024-06-26", - "capabilities": { - "vision": true, - "reasoning": true, - "functionCalling": true - }, - "thinkingOptions": { - "efforts": ["medium"] - }, - "modalities": { - "input": ["text", "image", "pdf"], - "output": ["text"] - } - }, "openai/o3-fast": { "displayName": "o3 (Fast)", "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "inputLimit": 100000, "maxOutputTokens": 100000, @@ -33452,7 +32587,6 @@ "displayName": "o3-mini", "description": "Smaller o-series reasoner for economical coding, math, and planning tasks", "lifecycle": "deprecated", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -33475,7 +32609,6 @@ "displayName": "o3 Pro", "description": "High-effort o3 tier for difficult technical reasoning and careful answers", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "inputLimit": 100000, "maxOutputTokens": 100000, @@ -33499,7 +32632,6 @@ "displayName": "o4-mini", "description": "Fast o-series model for compact reasoning, coding, and tool use", "lifecycle": "deprecated", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 100000, "knowledgeCutoff": "2024-05", @@ -33522,7 +32654,6 @@ "displayName": "o4-mini (Fast)", "description": "Fast o-series model for compact reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "inputLimit": 100000, "maxOutputTokens": 100000, @@ -33546,7 +32677,6 @@ "displayName": "text-embedding-3-large", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "inputLimit": 6656, "maxOutputTokens": 1536, @@ -33565,7 +32695,6 @@ "displayName": "text-embedding-3-small", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "inputLimit": 6656, "maxOutputTokens": 1536, @@ -33584,7 +32713,6 @@ "displayName": "text-embedding-ada-002", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "inputLimit": 6656, "maxOutputTokens": 1536, @@ -33603,7 +32731,6 @@ "displayName": "TTS-1", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2023-11-06", @@ -33621,7 +32748,6 @@ "displayName": "TTS-1 HD", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2023-11-06", @@ -33639,7 +32765,6 @@ "displayName": "Whisper", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2022-09-21", @@ -33657,7 +32782,6 @@ "displayName": "Embed v1 0.6b", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 0, "lastUpdated": "2026-02-26", @@ -33675,7 +32799,6 @@ "displayName": "Embed v1 4b", "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 0, "lastUpdated": "2026-02-26", @@ -33693,7 +32816,6 @@ "displayName": "Sonar", "description": "Sonar search model for current answers, retrieval, and citation-backed chat", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 127000, "maxOutputTokens": 8000, "knowledgeCutoff": "2025-02", @@ -33712,7 +32834,6 @@ "displayName": "Sonar Pro", "description": "Advanced Sonar search model for deeper research and cited synthesis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 8000, "knowledgeCutoff": "2025-09", @@ -33731,7 +32852,6 @@ "displayName": "Sonar Reasoning Pro", "description": "Web-grounded reasoning model for multi-step research and cited answers", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 127000, "maxOutputTokens": 8000, "knowledgeCutoff": "2025-09", @@ -33753,7 +32873,6 @@ "displayName": "Laguna S 2.1", "description": "Agentic coding model from Poolside in the XS size class for local deployment", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": false, @@ -33775,7 +32894,6 @@ "displayName": "Laguna S 2.1 Free", "description": "Free provider route for experiments, demos, and cost-sensitive chat workloads", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 32768, "structuredOutput": false, @@ -33798,7 +32916,6 @@ "displayName": "Flux Schnell", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2026-06-08", @@ -33816,7 +32933,6 @@ "displayName": "Arrow 1.1", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 131072, "maxOutputTokens": 131072, "lastUpdated": "2026-04-16", @@ -33834,7 +32950,6 @@ "displayName": "Recraft V2", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2024-03", @@ -33852,7 +32967,6 @@ "displayName": "Recraft V3", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 512, "maxOutputTokens": 0, "lastUpdated": "2024-10", @@ -33870,7 +32984,6 @@ "displayName": "Recraft V4", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-02-17", @@ -33888,7 +33001,6 @@ "displayName": "Recraft V4 Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-02-17", @@ -33906,7 +33018,6 @@ "displayName": "Recraft V4.1", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-05-14", @@ -33924,7 +33035,6 @@ "displayName": "Recraft V4.1 Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-05-14", @@ -33942,7 +33052,6 @@ "displayName": "Recraft V4.1 Utility", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-05-14", @@ -33960,7 +33069,6 @@ "displayName": "Recraft V4.1 Utility Pro", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-05-14", @@ -33978,7 +33086,6 @@ "displayName": "Fugu Ultra", "description": "Quality-first multi-agent model for hard research, analysis, and competitions", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 1000000, "structuredOutput": true, @@ -33997,7 +33104,6 @@ "displayName": "Sakana Namazu", "description": "Multi-agent model for routing expert agents across complex analytical tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "lastUpdated": "2026-08-03", @@ -34015,7 +33121,6 @@ "displayName": "Grok 4.1 Fast Non-Reasoning", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 1000000, "lastUpdated": "2025-11-19", @@ -34033,7 +33138,6 @@ "displayName": "Grok 4.1 Fast Reasoning", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 1000000, "lastUpdated": "2025-11-19", @@ -34051,7 +33155,6 @@ "displayName": "Grok 4.20 Multi-Agent", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 2000000, "maxOutputTokens": 2000000, "lastUpdated": "2026-03-10", @@ -34069,7 +33172,6 @@ "displayName": "Grok 4.20 Multi Agent Beta", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 2000000, "maxOutputTokens": 2000000, "lastUpdated": "2026-03-11", @@ -34087,7 +33189,6 @@ "displayName": "Grok 4.20 Non-Reasoning", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 2000000, "maxOutputTokens": 2000000, "lastUpdated": "2026-03-10", @@ -34105,7 +33206,6 @@ "displayName": "Grok 4.20 Beta Non-Reasoning", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 2000000, "maxOutputTokens": 2000000, "lastUpdated": "2026-03-11", @@ -34123,7 +33223,6 @@ "displayName": "Grok 4.20 Reasoning", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 2000000, "maxOutputTokens": 2000000, "lastUpdated": "2026-03-10", @@ -34141,7 +33240,6 @@ "displayName": "Grok 4.20 Beta Reasoning", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 2000000, "maxOutputTokens": 2000000, "lastUpdated": "2026-03-11", @@ -34159,15 +33257,18 @@ "displayName": "Grok 4.3", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 1000000, - "lastUpdated": "2026-04-30", + "structuredOutput": true, + "lastUpdated": "2026-04-17", "capabilities": { "vision": true, "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "efforts": ["low", "medium", "high"] + }, "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] @@ -34177,15 +33278,18 @@ "displayName": "Grok 4.5", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 500000, "maxOutputTokens": 500000, + "structuredOutput": true, "lastUpdated": "2026-07-08", "capabilities": { "vision": true, "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "efforts": ["low", "medium", "high"] + }, "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] @@ -34195,15 +33299,19 @@ "displayName": "Grok 4.6", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 500000, "maxOutputTokens": 500000, + "knowledgeCutoff": "2026-02-01", + "structuredOutput": true, "lastUpdated": "2026-08-12", "capabilities": { "vision": true, "reasoning": true, "functionCalling": true }, + "thinkingOptions": { + "efforts": ["low", "medium", "high"] + }, "modalities": { "input": ["text", "image"], "output": ["text"] @@ -34213,10 +33321,10 @@ "displayName": "Grok Build 0.1", "description": "Grok coding model for agentic engineering, edits, and codebase workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, - "lastUpdated": "2026-05-20", + "structuredOutput": true, + "lastUpdated": "2026-04-16", "capabilities": { "vision": true, "reasoning": true, @@ -34231,7 +33339,6 @@ "displayName": "Grok Imagine Image", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-01-28", @@ -34249,7 +33356,6 @@ "displayName": "Grok Imagine Image 2.0", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-08-07", @@ -34267,7 +33373,6 @@ "displayName": "Grok Imagine", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-01-28", @@ -34278,17 +33383,16 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "spacexai/grok-imagine-video-1.5": { "displayName": "Grok Imagine Video 1.5", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, - "lastUpdated": "2026-06-22", + "lastUpdated": "2026-05-30", "capabilities": { "vision": false, "reasoning": false, @@ -34296,14 +33400,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "spacexai/grok-imagine-video-1.5-preview": { "displayName": "Grok Imagine Video 1.5 Preview", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-05-30", @@ -34314,14 +33417,13 @@ }, "modalities": { "input": ["text"], - "output": [] + "output": ["video"] } }, "spacexai/grok-stt": { "displayName": "Grok STT", "description": "Speech transcription model for accurate audio-to-text and captioning workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-16", @@ -34339,7 +33441,6 @@ "displayName": "Grok TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-03-16", @@ -34357,7 +33458,6 @@ "displayName": "Grok Voice Think Fast 1.0", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-04-23", @@ -34375,7 +33475,6 @@ "displayName": "Grok Voice Think Fast 2.0", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 0, "maxOutputTokens": 0, "lastUpdated": "2026-07-29", @@ -34393,7 +33492,6 @@ "displayName": "StepFun 3.5 Flash", "description": "StepFun flash lane for quick multimodal reasoning and coding assistance", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 262114, "maxOutputTokens": 262114, "knowledgeCutoff": "2025-01", @@ -34415,7 +33513,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -34438,7 +33535,6 @@ "displayName": "Tencent Hy-MT2-Lite", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8000, "maxOutputTokens": 4000, "lastUpdated": "2026-06-12", @@ -34456,7 +33552,6 @@ "displayName": "Tencent Hy-MT2-Plus", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8000, "maxOutputTokens": 4000, "lastUpdated": "2026-06-12", @@ -34474,7 +33569,6 @@ "displayName": "Tencent Hy-MT2-Pro", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8000, "maxOutputTokens": 4000, "lastUpdated": "2026-05-21", @@ -34492,9 +33586,8 @@ "displayName": "Hy3", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", - "contextWindow": 256000, - "maxOutputTokens": 128000, + "contextWindow": 262144, + "maxOutputTokens": 262144, "lastUpdated": "2026-07-06", "capabilities": { "vision": false, @@ -34509,11 +33602,30 @@ "output": ["text"] } }, + "tencent/hy4-preview": { + "displayName": "Tencent Hy4 Preview", + "description": "A next-generation productivity model with significantly enhanced Agent and complex task execution capabilities.", + "lifecycle": "active", + "contextWindow": 1024000, + "maxOutputTokens": 64000, + "lastUpdated": "2026-08-28", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["none", "high"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "thinkingmachines/inkling": { "displayName": "Inkling", "description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 256000, "maxOutputTokens": 256000, "lastUpdated": "2026-07-15", @@ -34534,7 +33646,6 @@ "displayName": "Inkling Small", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 1000000, "lastUpdated": "2026-07-30", @@ -34555,7 +33666,6 @@ "displayName": "Voyage Rerank 2.5", "description": "Reranking model for improving retrieval quality in search and recommendation systems", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 32000, "lastUpdated": "2025-08-11", @@ -34573,7 +33683,6 @@ "displayName": "Voyage Rerank 2.5 Lite", "description": "Reranking model for improving retrieval quality in search and recommendation systems", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 32000, "lastUpdated": "2025-08-11", @@ -34591,7 +33700,6 @@ "displayName": "voyage-3-large", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-09", @@ -34609,7 +33717,6 @@ "displayName": "voyage-3.5", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2025-05-20", @@ -34627,7 +33734,6 @@ "displayName": "voyage-3.5-lite", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2025-05-20", @@ -34645,7 +33751,6 @@ "displayName": "voyage-4", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 0, "lastUpdated": "2026-03-06", @@ -34663,7 +33768,6 @@ "displayName": "voyage-4-large", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 0, "lastUpdated": "2026-03-06", @@ -34681,7 +33785,6 @@ "displayName": "voyage-4-lite", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 32000, "maxOutputTokens": 0, "lastUpdated": "2026-03-06", @@ -34699,7 +33802,6 @@ "displayName": "voyage-code-2", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-01", @@ -34717,7 +33819,6 @@ "displayName": "voyage-code-3", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-09", @@ -34735,7 +33836,6 @@ "displayName": "voyage-finance-2", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-03", @@ -34753,7 +33853,6 @@ "displayName": "voyage-law-2", "description": "General-purpose chat model for instruction following, writing, and analysis", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 8192, "maxOutputTokens": 1536, "lastUpdated": "2024-03", @@ -34771,7 +33870,6 @@ "displayName": "MiMo M2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "maxOutputTokens": 131100, "knowledgeCutoff": "2024-12", @@ -34793,7 +33891,6 @@ "displayName": "MiMo V2.5 Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1050000, "maxOutputTokens": 131000, "knowledgeCutoff": "2024-12", @@ -34811,11 +33908,31 @@ "output": ["text"] } }, + "xiaomi/mimo-v2.5-pro-ultraspeed": { + "displayName": "MiMo V2.5 Pro UltraSpeed", + "description": "MiMo pro model for strong multimodal reasoning and agent execution", + "lifecycle": "active", + "contextWindow": 1048576, + "maxOutputTokens": 131072, + "knowledgeCutoff": "2024-12", + "lastUpdated": "2026-06-09", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "toggle": true + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, "zai/glm-4.5": { "displayName": "GLM 4.5", "description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 96000, "knowledgeCutoff": "2025-07", @@ -34837,7 +33954,6 @@ "displayName": "GLM 4.5 Air", "description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 128000, "maxOutputTokens": 96000, "knowledgeCutoff": "2025-04", @@ -34859,7 +33975,6 @@ "displayName": "GLM 4.5V", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 66000, "maxOutputTokens": 16000, "knowledgeCutoff": "2025-08", @@ -34881,7 +33996,6 @@ "displayName": "GLM 4.6", "description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 96000, "knowledgeCutoff": "2025-04", @@ -34903,7 +34017,6 @@ "displayName": "GLM 4.7", "description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 120000, "knowledgeCutoff": "2025-04", @@ -34925,7 +34038,6 @@ "displayName": "GLM 4.7 Flash", "description": "Budget GLM lane for fast coding help, routing, and everyday automation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 131000, "knowledgeCutoff": "2025-04", @@ -34947,7 +34059,6 @@ "displayName": "GLM 4.7 FlashX", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-01", @@ -34969,7 +34080,6 @@ "displayName": "GLM-5", "description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 202800, "maxOutputTokens": 131100, "lastUpdated": "2026-02-12", @@ -34990,7 +34100,6 @@ "displayName": "GLM 5 Turbo", "description": "Faster GLM-5 lane for coding agents that need lower latency", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 202800, "maxOutputTokens": 131100, "structuredOutput": true, @@ -35012,7 +34121,6 @@ "displayName": "GLM 5.1", "description": "Strong GLM coding model for agentic engineering, terminals, and repository generation", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 202800, "maxOutputTokens": 64000, "structuredOutput": true, @@ -35034,7 +34142,6 @@ "displayName": "GLM 5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -35057,7 +34164,6 @@ "displayName": "GLM 5.2 Fast", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, "maxOutputTokens": 128000, "structuredOutput": true, @@ -35080,9 +34186,8 @@ "displayName": "GLM 5.3", "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 1000000, - "maxOutputTokens": 12800, + "maxOutputTokens": 1000000, "structuredOutput": true, "lastUpdated": "2026-08-14", "capabilities": { @@ -35098,11 +34203,31 @@ "output": ["text"] } }, + "zai/glm-5.3-flash": { + "displayName": "GLM 5.3 Flash", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131000, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image"], + "output": ["text"] + } + }, "zai/glm-5v-turbo": { "displayName": "GLM 5V Turbo", "description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway", "contextWindow": 200000, "maxOutputTokens": 128000, "lastUpdated": "2026-04-01", @@ -35125,7 +34250,6 @@ "displayName": "Grok 4.20 (Non-Reasoning)", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 1000000, "maxOutputTokens": 30000, "structuredOutput": true, @@ -35144,7 +34268,6 @@ "displayName": "Grok 4.20 (Reasoning)", "description": "Reasoning Grok for document-heavy analysis and long-horizon tool use", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 1000000, "maxOutputTokens": 30000, "structuredOutput": true, @@ -35163,7 +34286,6 @@ "displayName": "Grok 4.20 Multi-Agent", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 1000000, "maxOutputTokens": 30000, "structuredOutput": true, @@ -35185,7 +34307,6 @@ "displayName": "Grok 4.3", "description": "xAI's Grok for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 1000000, "maxOutputTokens": 30000, "structuredOutput": true, @@ -35207,7 +34328,6 @@ "displayName": "Grok 4.5", "description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 500000, "maxOutputTokens": 500000, "structuredOutput": true, @@ -35229,7 +34349,6 @@ "displayName": "Grok 4.6", "description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 500000, "maxOutputTokens": 500000, "knowledgeCutoff": "2026-02-01", @@ -35252,7 +34371,6 @@ "displayName": "Grok Build 0.1", "description": "Fast Grok coding model tuned for agentic engineering and iterative edits", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 256000, "maxOutputTokens": 256000, "structuredOutput": true, @@ -35271,7 +34389,6 @@ "displayName": "Grok Imagine Image", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 8000, "maxOutputTokens": 0, "lastUpdated": "2026-01-28", @@ -35282,14 +34399,13 @@ }, "modalities": { "input": ["text", "image", "pdf"], - "output": ["image"] + "output": ["image", "pdf"] } }, "grok-imagine-image-2.0": { "displayName": "Grok Imagine Image 2.0", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 8000, "maxOutputTokens": 0, "lastUpdated": "2026-08-07", @@ -35300,14 +34416,13 @@ }, "modalities": { "input": ["text", "image", "pdf"], - "output": ["image"] + "output": ["image", "pdf"] } }, "grok-imagine-image-quality": { "displayName": "Grok Imagine Image Quality", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 8000, "maxOutputTokens": 0, "lastUpdated": "2026-04-03", @@ -35318,14 +34433,13 @@ }, "modalities": { "input": ["text", "image", "pdf"], - "output": ["image"] + "output": ["image", "pdf"] } }, "grok-imagine-video": { "displayName": "Grok Imagine Video", "description": "Image model for prompt-driven generation, editing, and visual design workflows", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 1024, "maxOutputTokens": 0, "lastUpdated": "2026-01-28", @@ -35335,15 +34449,14 @@ "functionCalling": false }, "modalities": { - "input": ["text", "image", "pdf"], - "output": [] + "input": ["text", "image", "video", "pdf"], + "output": ["video"] } }, "grok-imagine-video-1.5": { "displayName": "Grok Imagine Video 1.5", "description": "Video model for image-to-video generation, editing, and extension workflows", "lifecycle": "active", - "docsUrl": "https://docs.x.ai/docs/models", "contextWindow": 1024, "maxOutputTokens": 0, "lastUpdated": "2026-05-30", @@ -35354,7 +34467,7 @@ }, "modalities": { "input": ["text", "image", "audio", "pdf"], - "output": [] + "output": ["video"] } } }, @@ -35363,7 +34476,6 @@ "displayName": "MiMo-V2-Flash", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-12-01", @@ -35385,7 +34497,6 @@ "displayName": "MiMo-V2-Omni", "description": "Legacy model retained for compatibility with older integrations", "lifecycle": "deprecated", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 262144, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35399,7 +34510,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -35407,7 +34518,6 @@ "displayName": "MiMo-V2-Pro", "description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35429,7 +34539,6 @@ "displayName": "MiMo-V2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35443,7 +34552,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -35451,7 +34560,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35473,7 +34581,6 @@ "displayName": "MiMo-V2.5-Pro-UltraSpeed", "description": "MiMo pro model for strong multimodal reasoning and agent execution", "lifecycle": "beta", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35497,7 +34604,6 @@ "displayName": "MiMo-V2-Pro", "description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35520,7 +34626,6 @@ "displayName": "MiMo-V2-TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-03-18", @@ -35539,7 +34644,6 @@ "displayName": "MiMo-V2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35554,7 +34658,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -35562,7 +34666,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35585,7 +34688,6 @@ "displayName": "MiMo-V2.5-TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35604,7 +34706,6 @@ "displayName": "MiMo-V2.5-TTS-VoiceClone", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35623,7 +34724,6 @@ "displayName": "MiMo-V2.5-TTS-VoiceDesign", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35644,7 +34744,6 @@ "displayName": "MiMo-V2-Pro", "description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35667,7 +34766,6 @@ "displayName": "MiMo-V2-TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-03-18", @@ -35686,7 +34784,6 @@ "displayName": "MiMo-V2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35701,7 +34798,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -35709,7 +34806,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35732,7 +34828,6 @@ "displayName": "MiMo-V2.5-TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35751,7 +34846,6 @@ "displayName": "MiMo-V2.5-TTS-VoiceClone", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35770,7 +34864,6 @@ "displayName": "MiMo-V2.5-TTS-VoiceDesign", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35791,7 +34884,6 @@ "displayName": "MiMo-V2-Pro", "description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks", "lifecycle": "deprecated", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35814,7 +34906,6 @@ "displayName": "MiMo-V2-TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-03-18", @@ -35833,7 +34924,6 @@ "displayName": "MiMo-V2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35848,7 +34938,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -35856,7 +34946,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -35879,7 +34968,6 @@ "displayName": "MiMo-V2.5-TTS", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35898,7 +34986,6 @@ "displayName": "MiMo-V2.5-TTS-VoiceClone", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35917,7 +35004,6 @@ "displayName": "MiMo-V2.5-TTS-VoiceDesign", "description": "Speech generation model for controllable voice, narration, and audio delivery", "lifecycle": "active", - "docsUrl": "https://platform.xiaomimimo.com/#/docs", "contextWindow": 8192, "maxOutputTokens": 8192, "lastUpdated": "2026-04-22", @@ -35938,7 +35024,6 @@ "displayName": "GLM-4.5", "description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 131072, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-04", @@ -35960,7 +35045,6 @@ "displayName": "GLM-4.5-Air", "description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 131072, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-04", @@ -35982,7 +35066,6 @@ "displayName": "GLM-4.5-Flash", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 131072, "maxOutputTokens": 98304, "knowledgeCutoff": "2025-04", @@ -36005,7 +35088,6 @@ "displayName": "GLM-4.5V", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 64000, "maxOutputTokens": 16384, "knowledgeCutoff": "2025-04", @@ -36019,7 +35101,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -36027,7 +35109,6 @@ "displayName": "GLM-4.6", "description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -36049,7 +35130,6 @@ "displayName": "GLM-4.6V", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 128000, "maxOutputTokens": 32768, "knowledgeCutoff": "2025-04", @@ -36063,7 +35143,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -36071,7 +35151,6 @@ "displayName": "GLM-4.7", "description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -36093,7 +35172,6 @@ "displayName": "GLM-4.7-Flash", "description": "Budget GLM lane for fast coding help, routing, and everyday automation", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 200000, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -36116,7 +35194,6 @@ "displayName": "GLM-4.7-FlashX", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 200000, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -36138,7 +35215,6 @@ "displayName": "GLM-5", "description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 204800, "maxOutputTokens": 131072, "lastUpdated": "2026-02-12", @@ -36159,7 +35235,6 @@ "displayName": "GLM-5-Turbo", "description": "Faster GLM-5 lane for coding agents that need lower latency", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 200000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -36181,7 +35256,6 @@ "displayName": "GLM-5.1", "description": "Strong GLM coding model for agentic engineering, terminals, and repository generation", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 200000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -36203,7 +35277,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -36221,11 +35294,52 @@ "output": ["text"] } }, + "glm-5.3": { + "displayName": "GLM-5.3", + "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "glm-5.3-flash": { + "displayName": "GLM-5.3-Flash", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "video", "pdf"], + "output": ["text"] + } + }, "glm-5v-turbo": { "displayName": "GLM-5V-Turbo", "description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/guides/overview/pricing", "contextWindow": 200000, "maxOutputTokens": 131072, "lastUpdated": "2026-04-01", @@ -36238,7 +35352,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } } @@ -36248,7 +35362,6 @@ "displayName": "GLM-4.7", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/devpack/overview", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-04", @@ -36271,7 +35384,6 @@ "displayName": "GLM-5-Turbo", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/devpack/overview", "contextWindow": 200000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -36294,7 +35406,6 @@ "displayName": "GLM-5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/devpack/overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -36317,7 +35428,6 @@ "displayName": "GLM-5.2 Highspeed", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/devpack/overview", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -36340,7 +35450,50 @@ "displayName": "GLM-5.3", "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", "lifecycle": "active", - "docsUrl": "https://docs.z.ai/devpack/overview", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-14", + "isFree": true, + "capabilities": { + "vision": false, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text"], + "output": ["text"] + } + }, + "glm-5.3-flash": { + "displayName": "GLM-5.3-Flash", + "description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks", + "lifecycle": "active", + "contextWindow": 1000000, + "maxOutputTokens": 131072, + "structuredOutput": true, + "lastUpdated": "2026-08-26", + "isFree": true, + "capabilities": { + "vision": true, + "reasoning": true, + "functionCalling": true + }, + "thinkingOptions": { + "efforts": ["low", "high", "max"] + }, + "modalities": { + "input": ["text", "image", "video", "pdf"], + "output": ["text"] + } + }, + "glm-5.3-highspeed": { + "displayName": "GLM-5.3 Highspeed", + "description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery", + "lifecycle": "active", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -36365,7 +35518,6 @@ "displayName": "Claude 3.5 Haiku", "description": "Fast Claude model for responsive assistance, classification, and lightweight agents", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36384,7 +35536,6 @@ "displayName": "Claude 3.7 Sonnet", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36406,7 +35557,6 @@ "displayName": "Claude Fable 5", "description": "Claude model for creative writing, analysis, and controlled agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -36428,7 +35578,6 @@ "displayName": "Claude Haiku 4.5", "description": "Fast Claude model for responsive assistance, classification, and lightweight agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36447,7 +35596,6 @@ "displayName": "Claude Opus 4", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-01-01", @@ -36469,7 +35617,6 @@ "displayName": "Claude Opus 4.1", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36491,7 +35638,6 @@ "displayName": "Claude Opus 4.5", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36513,7 +35659,6 @@ "displayName": "Claude Opus 4.6", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-05-31", @@ -36535,7 +35680,6 @@ "displayName": "Claude Opus 4.7", "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -36557,7 +35701,6 @@ "displayName": "Claude Opus 4.8", "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01", @@ -36579,7 +35722,6 @@ "displayName": "Claude Sonnet 4", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36601,7 +35743,6 @@ "displayName": "Claude Sonnet 4.5", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36623,7 +35764,6 @@ "displayName": "Claude Sonnet 4.6", "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-08-31", @@ -36645,7 +35785,6 @@ "displayName": "Claude Sonnet 5", "description": "Everyday Claude agent model for coding, planning, browsing, and general work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -36667,7 +35806,6 @@ "displayName": "Claude Sonnet 5 (Free)", "description": "Everyday Claude agent model for coding, planning, browsing, and general work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 128000, "knowledgeCutoff": "2026-01-31", @@ -36690,7 +35828,6 @@ "displayName": "ERNIE 5.0", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36701,7 +35838,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -36709,7 +35846,6 @@ "displayName": "DeepSeek-V3.2 (Non-thinking Mode)", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36728,7 +35864,6 @@ "displayName": "DeepSeek V3.2", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36750,7 +35885,6 @@ "displayName": "DeepSeek-V3.2-Exp", "description": "DeepSeek chat model for instruction following, coding, and analysis", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 163000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36772,7 +35906,6 @@ "displayName": "DeepSeek V4 Flash", "description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -36796,7 +35929,6 @@ "displayName": "DeepSeek V4 Pro", "description": "Open MoE flagship with million-token context for coding and long agent runs", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 384000, "knowledgeCutoff": "2025-05", @@ -36820,7 +35952,6 @@ "displayName": "Gemini 2.5 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36842,7 +35973,6 @@ "displayName": "Gemini 2.5 Flash Lite", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36861,7 +35991,6 @@ "displayName": "Gemini 2.5 Pro", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36875,7 +36004,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["pdf", "image", "text", "audio"], + "input": ["pdf", "image", "text", "audio", "video"], "output": ["text"] } }, @@ -36883,7 +36012,6 @@ "displayName": "Gemini 3 Flash Preview", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -36905,7 +36033,6 @@ "displayName": "Gemini 3.1 Flash Lite", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -36920,7 +36047,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -36928,7 +36055,6 @@ "displayName": "Gemini 3.1 Flash Lite Preview", "description": "Low-latency Gemini model for high-volume multimodal and agent workloads", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "maxOutputTokens": 65530, "lastUpdated": "2025-03-20", @@ -36938,7 +36064,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -36946,7 +36072,6 @@ "displayName": "Gemini 3.1 Pro Preview", "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048000, "maxOutputTokens": 64000, "knowledgeCutoff": "2026-02-19", @@ -36960,7 +36085,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "pdf", "audio"], + "input": ["text", "image", "pdf", "audio", "video"], "output": ["text"] } }, @@ -36968,7 +36093,6 @@ "displayName": "Gemini 3.5 Flash", "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048576, "maxOutputTokens": 65536, "knowledgeCutoff": "2025-01", @@ -36983,7 +36107,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "video", "audio", "pdf"], "output": ["text"] } }, @@ -36991,7 +36115,6 @@ "displayName": "Ling-1T", "description": "Tool-capable chat model for instruction following and agentic application workflows", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37010,7 +36133,6 @@ "displayName": "Ring-1T", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37029,7 +36151,6 @@ "displayName": "inclusionAI: Ring-2.6-1T", "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262000, "maxOutputTokens": 65000, "knowledgeCutoff": "2025-12-31", @@ -37048,7 +36169,6 @@ "displayName": "KAT-Coder-Pro-V2", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 80000, "lastUpdated": "2026-03-30", @@ -37066,7 +36186,6 @@ "displayName": "MiniMax M2", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 204000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37085,7 +36204,6 @@ "displayName": "MiniMax M2.1", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 204000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37104,7 +36222,6 @@ "displayName": "MiniMax M2.5", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01-01", @@ -37123,7 +36240,6 @@ "displayName": "MiniMax M2.5 highspeed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 204800, "maxOutputTokens": 131072, "knowledgeCutoff": "2025-01-01", @@ -37142,7 +36258,6 @@ "displayName": "MiniMax M2.7", "description": "MiniMax model for chat, coding, office work, and agentic tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 204800, "maxOutputTokens": 131070, "knowledgeCutoff": "2025-01-01", @@ -37161,7 +36276,6 @@ "displayName": "MiniMax M2.7 highspeed", "description": "High-speed MiniMax model for low-latency coding and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 204800, "maxOutputTokens": 131070, "knowledgeCutoff": "2025-01-01", @@ -37180,9 +36294,8 @@ "displayName": "MiniMax-M3", "description": "MiniMax multimodal model for long-context coding, perception, and agent planning", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", - "contextWindow": 512000, - "maxOutputTokens": 128000, + "contextWindow": 1048576, + "maxOutputTokens": 512000, "lastUpdated": "2026-06-01", "capabilities": { "vision": true, @@ -37193,7 +36306,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37201,7 +36314,6 @@ "displayName": "Kimi K2 0905", "description": "Kimi model for long-context chat, coding, and agentic reasoning", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37220,7 +36332,6 @@ "displayName": "Kimi K2 Thinking", "description": "Kimi reasoning model for long-horizon research, planning, and tool use", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37239,7 +36350,6 @@ "displayName": "Kimi K2 Thinking Turbo", "description": "Kimi reasoning model for long-horizon research, planning, and tool use", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37258,7 +36368,6 @@ "displayName": "Kimi K2.5", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37272,7 +36381,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37280,7 +36389,6 @@ "displayName": "Kimi K2.6", "description": "Kimi multimodal agent model for visual understanding, coding, and planning", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262140, "maxOutputTokens": 262140, "knowledgeCutoff": "2025-01-01", @@ -37294,7 +36402,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37302,7 +36410,6 @@ "displayName": "Kimi K2.7 Code", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -37314,7 +36421,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37322,7 +36429,6 @@ "displayName": "Kimi K2.7 Code (Free)", "description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262144, "maxOutputTokens": 262144, "knowledgeCutoff": "2025-01", @@ -37335,7 +36441,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37343,7 +36449,6 @@ "displayName": "Kimi K3", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -37358,7 +36463,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37366,7 +36471,6 @@ "displayName": "Kimi K3 (Free)", "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048576, "maxOutputTokens": 131072, "structuredOutput": true, @@ -37382,7 +36486,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37390,7 +36494,6 @@ "displayName": "GPT-5", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37412,7 +36515,6 @@ "displayName": "GPT-5 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37434,7 +36536,6 @@ "displayName": "GPT-5.1", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37456,7 +36557,6 @@ "displayName": "GPT-5.1 Chat", "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37475,7 +36575,6 @@ "displayName": "GPT-5.1-Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37497,7 +36596,6 @@ "displayName": "GPT-5.1-Codex-Mini", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37519,7 +36617,6 @@ "displayName": "GPT-5.2", "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37541,7 +36638,6 @@ "displayName": "GPT-5.2-Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37563,7 +36659,6 @@ "displayName": "GPT-5.2-Pro", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -37585,7 +36680,6 @@ "displayName": "GPT-5.3 Chat", "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 16380, "knowledgeCutoff": "2025-08-31", @@ -37604,7 +36698,6 @@ "displayName": "GPT-5.3 Codex", "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -37626,7 +36719,6 @@ "displayName": "GPT-5.4", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -37648,7 +36740,6 @@ "displayName": "GPT-5.4 Mini", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -37667,7 +36758,6 @@ "displayName": "GPT-5.4 Nano", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -37686,7 +36776,6 @@ "displayName": "GPT-5.4 Pro", "description": "Frontier GPT model for professional reasoning, coding, and multimodal work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-08-31", @@ -37708,7 +36797,6 @@ "displayName": "GPT-5.5", "description": "Default frontier GPT for coding, computer use, research, and knowledge work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -37732,7 +36820,6 @@ "displayName": "GPT-5.5 Instant", "description": "Compact GPT model for low-latency assistance and high-volume workloads", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 400000, "inputLimit": 400000, "maxOutputTokens": 128000, @@ -37756,7 +36843,6 @@ "displayName": "GPT-5.5 Pro", "description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -37780,7 +36866,6 @@ "displayName": "GPT-5.6 Luna", "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -37804,7 +36889,6 @@ "displayName": "GPT-5.6 Sol", "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -37828,7 +36912,6 @@ "displayName": "GPT-5.6 Terra", "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1050000, "inputLimit": 922000, "maxOutputTokens": 128000, @@ -37852,7 +36935,6 @@ "displayName": "Qwen3-Coder-Plus", "description": "Qwen coding model for software agents, repository edits, and code reasoning", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37871,7 +36953,6 @@ "displayName": "Qwen3-Max-Thinking", "description": "Qwen reasoning model for deliberate problem solving, math, and coding", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37890,7 +36971,6 @@ "displayName": "Qwen3.5 Flash", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1020000, "maxOutputTokens": 1020000, "knowledgeCutoff": "2025-01-01", @@ -37909,7 +36989,6 @@ "displayName": "Qwen3.5 Plus", "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -37931,7 +37010,6 @@ "displayName": "Qwen3.6-Plus", "description": "Qwen instruction model for multilingual chat, reasoning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 64000, "lastUpdated": "2026-03-30", @@ -37952,7 +37030,6 @@ "displayName": "Qwen3.7 Max", "description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 65536, "lastUpdated": "2026-05-21", @@ -37973,7 +37050,6 @@ "displayName": "Qwen3.7 Plus", "description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-04", @@ -37987,7 +37063,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -37995,7 +37071,6 @@ "displayName": "Agnes 1.5 Lite", "description": "Efficient model for low-latency assistance, extraction, and routine automation", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 256000, "lastUpdated": "2026-03-26", @@ -38013,7 +37088,6 @@ "displayName": "Agnes 1.5 Pro", "description": "Flagship model for demanding analysis, coding, and production agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 256000, "lastUpdated": "2026-03-21", @@ -38031,7 +37105,6 @@ "displayName": "Step-3", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 65536, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38050,7 +37123,6 @@ "displayName": "Step 3.5 Flash", "description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38069,7 +37141,6 @@ "displayName": "Step 3.7 Flash", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -38084,7 +37155,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38092,7 +37163,6 @@ "displayName": "Step 3.7 Flash (Free)", "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "inputLimit": 256000, "maxOutputTokens": 256000, @@ -38108,7 +37178,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38116,7 +37186,6 @@ "displayName": "Hy3 preview", "description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "lastUpdated": "2026-04-20", @@ -38137,7 +37206,6 @@ "displayName": "Doubao-Seed-1.8", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38151,7 +37219,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38159,7 +37227,6 @@ "displayName": "Doubao Seed 2.0 Code", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 32000, "knowledgeCutoff": "2025-01-01", @@ -38178,7 +37245,6 @@ "displayName": "Doubao-Seed-2.0-lite", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2026-02-14", @@ -38192,7 +37258,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38200,7 +37266,6 @@ "displayName": "Doubao-Seed-2.0-mini", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2026-02-14", @@ -38214,7 +37279,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38222,7 +37287,6 @@ "displayName": "Doubao-Seed-2.0-pro", "description": "Multimodal reasoning model for visual analysis, planning, and tool use", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2026-02-14", @@ -38236,7 +37300,7 @@ "efforts": ["low", "medium", "high"] }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38244,7 +37308,6 @@ "displayName": "Doubao-Seed-Code", "description": "Coding model for repository understanding, refactors, and agentic engineering tasks", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38266,7 +37329,6 @@ "displayName": "Grok 4", "description": "Grok model for agentic tool use, reasoning, coding, and live assistance", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38285,7 +37347,6 @@ "displayName": "Grok 4 Fast", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 2000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38307,7 +37368,6 @@ "displayName": "Grok 4.1 Fast", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 2000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38329,7 +37389,6 @@ "displayName": "Grok 4.1 Fast Non Reasoning", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 2000000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38348,7 +37407,6 @@ "displayName": "Grok 4.2 Fast", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 2000000, "maxOutputTokens": 30000, "knowledgeCutoff": "2025-08-31", @@ -38359,7 +37417,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38367,7 +37425,6 @@ "displayName": "Grok 4.2 Fast Non Reasoning", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 2000000, "maxOutputTokens": 30000, "knowledgeCutoff": "2025-08-31", @@ -38378,7 +37435,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38386,7 +37443,6 @@ "displayName": "Grok 4.3", "description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 1000000, "structuredOutput": true, @@ -38408,7 +37464,6 @@ "displayName": "Grok 4.5", "description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 500000, "maxOutputTokens": 500000, "structuredOutput": true, @@ -38430,7 +37485,6 @@ "displayName": "Grok Build 0.1", "description": "Fast Grok coding model tuned for agentic engineering and iterative edits", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 256000, "structuredOutput": true, @@ -38449,7 +37503,6 @@ "displayName": "Grok Code Fast 1", "description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work", "lifecycle": "deprecated", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 256000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38468,7 +37521,6 @@ "displayName": "MiMo-V2-Flash", "description": "MiMo flash model for fast multimodal assistance and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 262144, "maxOutputTokens": 65536, "knowledgeCutoff": "2024-12-01", @@ -38490,7 +37542,6 @@ "displayName": "MiMo V2 Omni", "description": "MiMo omni model for text, image, video, audio, and agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 265000, "maxOutputTokens": 265000, "knowledgeCutoff": "2024-12", @@ -38504,7 +37555,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio", "pdf"], + "input": ["text", "image", "audio", "video", "pdf"], "output": ["text"] } }, @@ -38512,7 +37563,6 @@ "displayName": "MiMo V2 Pro", "description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 256000, "knowledgeCutoff": "2024-12", @@ -38534,7 +37584,6 @@ "displayName": "MiMo-V2.5", "description": "Open MiMo model for multimodal coding agents and long-context automation", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -38548,7 +37597,7 @@ "toggle": true }, "modalities": { - "input": ["text", "image", "audio"], + "input": ["text", "image", "audio", "video"], "output": ["text"] } }, @@ -38556,7 +37605,6 @@ "displayName": "MiMo-V2.5-Pro", "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1048576, "maxOutputTokens": 131072, "knowledgeCutoff": "2024-12", @@ -38578,7 +37626,6 @@ "displayName": "GLM 4.5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38600,7 +37647,6 @@ "displayName": "GLM 4.5 Air", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 128000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38622,7 +37668,6 @@ "displayName": "GLM 4.6", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38641,7 +37686,6 @@ "displayName": "GLM 4.6V", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38652,7 +37696,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38660,7 +37704,6 @@ "displayName": "GLM 4.6V FlashX", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38671,7 +37714,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38679,7 +37722,6 @@ "displayName": "GLM 4.6V Flash (Free)", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38691,7 +37733,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image"], + "input": ["text", "image", "video"], "output": ["text"] } }, @@ -38699,7 +37741,6 @@ "displayName": "GLM 4.7", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38718,7 +37759,6 @@ "displayName": "GLM 4.7 Flash (Free)", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38738,7 +37778,6 @@ "displayName": "GLM 4.7 FlashX", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 64000, "knowledgeCutoff": "2025-01-01", @@ -38757,7 +37796,6 @@ "displayName": "GLM 5", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-01-01", @@ -38776,7 +37814,6 @@ "displayName": "GLM 5 Turbo", "description": "Efficient GLM model for fast reasoning, coding, and agent workflows", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 128000, "knowledgeCutoff": "2025-01-01", @@ -38795,7 +37832,6 @@ "displayName": "GLM-5.1", "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -38817,7 +37853,6 @@ "displayName": "GLM 5.2", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -38839,7 +37874,6 @@ "displayName": "GLM 5.2 (Free)", "description": "Open flagship GLM for long-horizon coding agents and million-token context work", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 1000000, "maxOutputTokens": 131072, "structuredOutput": true, @@ -38862,7 +37896,6 @@ "displayName": "GLM 5V Turbo", "description": "GLM vision model for visual reasoning, documents, and multimodal agents", "lifecycle": "active", - "docsUrl": "https://docs.zenmux.ai", "contextWindow": 200000, "maxOutputTokens": 128000, "lastUpdated": "2026-04-01", @@ -38872,7 +37905,7 @@ "functionCalling": true }, "modalities": { - "input": ["text", "image", "pdf"], + "input": ["text", "image", "video", "pdf"], "output": ["text"] } } @@ -39108,16 +38141,6 @@ "inputUsdPer1M": 0.035, "outputUsdPer1M": 0.035 }, - { - "modelKey": "alibaba:qwen3-coder-30b-a3b-instruct", - "inputUsdPer1M": 0.45, - "outputUsdPer1M": 2.25 - }, - { - "modelKey": "alibaba:qwen3-coder-480b-a35b-instruct", - "inputUsdPer1M": 1.5, - "outputUsdPer1M": 7.5 - }, { "modelKey": "alibaba:qwen3-coder-flash", "inputUsdPer1M": 0.3, @@ -39228,6 +38251,13 @@ "cacheReadUsdPer1M": 0.5, "cacheWriteUsdPer1M": 3.125 }, + { + "modelKey": "alibaba:qwen3.8-flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.47, + "cacheReadUsdPer1M": 0.016, + "cacheWriteUsdPer1M": 0.2 + }, { "modelKey": "alibaba:qwen3.8-max", "inputUsdPer1M": 2, @@ -39307,17 +38337,6 @@ "outputUsdPer1M": 0.87, "cacheReadUsdPer1M": 0.003625 }, - { - "modelKey": "alibaba-cn:glm-5", - "inputUsdPer1M": 0.86, - "outputUsdPer1M": 3.15 - }, - { - "modelKey": "alibaba-cn:glm-5.1", - "inputUsdPer1M": 0.87, - "outputUsdPer1M": 3.48, - "cacheReadUsdPer1M": 0.17 - }, { "modelKey": "alibaba-cn:glm-5.2", "inputUsdPer1M": 1.1, @@ -39426,7 +38445,9 @@ { "modelKey": "alibaba-cn:qwen-plus", "inputUsdPer1M": 0.115, - "outputUsdPer1M": 0.287 + "outputUsdPer1M": 0.287, + "cacheReadUsdPer1M": 0.012, + "cacheWriteUsdPer1M": 0.144 }, { "modelKey": "alibaba-cn:qwen-plus-character", @@ -39533,16 +38554,6 @@ "inputUsdPer1M": 0.032, "outputUsdPer1M": 0.032 }, - { - "modelKey": "alibaba-cn:qwen3-coder-30b-a3b-instruct", - "inputUsdPer1M": 0.216, - "outputUsdPer1M": 0.861 - }, - { - "modelKey": "alibaba-cn:qwen3-coder-480b-a35b-instruct", - "inputUsdPer1M": 0.861, - "outputUsdPer1M": 3.441 - }, { "modelKey": "alibaba-cn:qwen3-coder-flash", "inputUsdPer1M": 0.144, @@ -39593,11 +38604,6 @@ "inputUsdPer1M": 0.143353, "outputUsdPer1M": 1.433525 }, - { - "modelKey": "alibaba-cn:qwen3.5-397b-a17b", - "inputUsdPer1M": 0.43, - "outputUsdPer1M": 2.58 - }, { "modelKey": "alibaba-cn:qwen3.5-flash", "inputUsdPer1M": 0.172, @@ -39627,6 +38633,13 @@ "cacheReadUsdPer1M": 0.5, "cacheWriteUsdPer1M": 3.125 }, + { + "modelKey": "alibaba-cn:qwen3.8-flash", + "inputUsdPer1M": 0.11875, + "outputUsdPer1M": 0.40073, + "cacheReadUsdPer1M": 0.01187, + "cacheWriteUsdPer1M": 0.14844 + }, { "modelKey": "alibaba-cn:qwen3.8-max", "inputUsdPer1M": 1.77744, @@ -39860,6 +38873,18 @@ "outputUsdPer1M": 4.4, "cacheReadUsdPer1M": 0.26 }, + { + "modelKey": "cloudflare-workers-ai:@cf/zai-org/glm-5.3", + "inputUsdPer1M": 1.4, + "outputUsdPer1M": 4.4, + "cacheReadUsdPer1M": 0.26 + }, + { + "modelKey": "cloudflare-workers-ai:@cf/zai-org/glm-5.3-flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.5, + "cacheReadUsdPer1M": 0.03 + }, { "modelKey": "deepinfra:deepseek-ai/DeepSeek-R1-0528", "inputUsdPer1M": 0.5, @@ -40171,19 +39196,25 @@ "cacheReadUsdPer1M": 0.14 }, { - "modelKey": "deepseek:deepseek-chat", - "inputUsdPer1M": 0.14, - "outputUsdPer1M": 0.28, - "cacheReadUsdPer1M": 0.0028 + "modelKey": "deepinfra:zai-org/GLM-5.3", + "inputUsdPer1M": 1.2, + "outputUsdPer1M": 4, + "cacheReadUsdPer1M": 0.12 + }, + { + "modelKey": "deepinfra:zai-org/GLM-5.3-Flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.5, + "cacheReadUsdPer1M": 0.03 }, { - "modelKey": "deepseek:deepseek-reasoner", + "modelKey": "deepseek:deepseek-v4-flash", "inputUsdPer1M": 0.14, "outputUsdPer1M": 0.28, "cacheReadUsdPer1M": 0.0028 }, { - "modelKey": "deepseek:deepseek-v4-flash", + "modelKey": "deepseek:deepseek-v4-flash-vision-exp", "inputUsdPer1M": 0.14, "outputUsdPer1M": 0.28, "cacheReadUsdPer1M": 0.0028 @@ -40194,24 +39225,12 @@ "outputUsdPer1M": 0.87, "cacheReadUsdPer1M": 0.003625 }, - { - "modelKey": "fireworks-ai:accounts/fireworks/models/deepseek-v4-flash", - "inputUsdPer1M": 0.14, - "outputUsdPer1M": 0.28, - "cacheReadUsdPer1M": 0.028 - }, { "modelKey": "fireworks-ai:accounts/fireworks/models/deepseek-v4-flash-0731", "inputUsdPer1M": 0.14, "outputUsdPer1M": 0.28, "cacheReadUsdPer1M": 0.028 }, - { - "modelKey": "fireworks-ai:accounts/fireworks/models/deepseek-v4-pro", - "inputUsdPer1M": 1.74, - "outputUsdPer1M": 3.48, - "cacheReadUsdPer1M": 0.145 - }, { "modelKey": "fireworks-ai:accounts/fireworks/models/deepseek-v4-pro-0813", "inputUsdPer1M": 1.32, @@ -40224,18 +39243,24 @@ "outputUsdPer1M": 4.4, "cacheReadUsdPer1M": 0.14 }, + { + "modelKey": "fireworks-ai:accounts/fireworks/models/glm-5p3", + "inputUsdPer1M": 1.4, + "outputUsdPer1M": 4.4, + "cacheReadUsdPer1M": 0.26 + }, + { + "modelKey": "fireworks-ai:accounts/fireworks/models/glm-5p3-flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.5, + "cacheReadUsdPer1M": 0.029 + }, { "modelKey": "fireworks-ai:accounts/fireworks/models/gpt-oss-120b", "inputUsdPer1M": 0.15, "outputUsdPer1M": 0.6, "cacheReadUsdPer1M": 0.015 }, - { - "modelKey": "fireworks-ai:accounts/fireworks/models/gpt-oss-20b", - "inputUsdPer1M": 0.07, - "outputUsdPer1M": 0.3, - "cacheReadUsdPer1M": 0.035 - }, { "modelKey": "fireworks-ai:accounts/fireworks/models/inkling", "inputUsdPer1M": 1, @@ -40260,12 +39285,6 @@ "outputUsdPer1M": 15, "cacheReadUsdPer1M": 0.3 }, - { - "modelKey": "fireworks-ai:accounts/fireworks/models/minimax-m2p7", - "inputUsdPer1M": 0.3, - "outputUsdPer1M": 1.2, - "cacheReadUsdPer1M": 0.06 - }, { "modelKey": "fireworks-ai:accounts/fireworks/models/minimax-m3", "inputUsdPer1M": 0.3, @@ -40308,24 +39327,6 @@ "outputUsdPer1M": 6.6, "cacheReadUsdPer1M": 0.21 }, - { - "modelKey": "fireworks-ai:accounts/fireworks/routers/kimi-k2p6-fast", - "inputUsdPer1M": 2, - "outputUsdPer1M": 8, - "cacheReadUsdPer1M": 0.3 - }, - { - "modelKey": "fireworks-ai:accounts/fireworks/routers/kimi-k2p6-turbo", - "inputUsdPer1M": 2, - "outputUsdPer1M": 8, - "cacheReadUsdPer1M": 0.3 - }, - { - "modelKey": "fireworks-ai:accounts/fireworks/routers/kimi-k2p7-code-fast", - "inputUsdPer1M": 1.9, - "outputUsdPer1M": 8, - "cacheReadUsdPer1M": 0.38 - }, { "modelKey": "fireworks-ai:accounts/fireworks/routers/kimi-k3-fast", "inputUsdPer1M": 4.5, @@ -40469,11 +39470,6 @@ "inputUsdPer1M": 1.5, "outputUsdPer1M": 17.5 }, - { - "modelKey": "google:gemini-robotics-er-1.6-preview", - "inputUsdPer1M": 1, - "outputUsdPer1M": 5 - }, { "modelKey": "google:lyria-3-clip-preview", "inputUsdPer1M": 0, @@ -40532,6 +39528,11 @@ "outputUsdPer1M": 3, "cacheReadUsdPer1M": 0.3 }, + { + "modelKey": "groq:qwen/qwen3.8-27b", + "inputUsdPer1M": 0.8, + "outputUsdPer1M": 4 + }, { "modelKey": "huggingface:deepseek-ai/DeepSeek-R1", "inputUsdPer1M": 0.7, @@ -40793,6 +39794,11 @@ "inputUsdPer1M": 2.5, "outputUsdPer1M": 6.25 }, + { + "modelKey": "huggingface:Qwen/Qwen3.8-27B", + "inputUsdPer1M": 0.4, + "outputUsdPer1M": 3 + }, { "modelKey": "huggingface:stepfun-ai/Step-3.5-Flash", "inputUsdPer1M": 0.1, @@ -40886,6 +39892,16 @@ "inputUsdPer1M": 1.4, "outputUsdPer1M": 4.4 }, + { + "modelKey": "huggingface:zai-org/GLM-5.3", + "inputUsdPer1M": 1.4, + "outputUsdPer1M": 4.4 + }, + { + "modelKey": "huggingface:zai-org/GLM-5.3-Flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.5 + }, { "modelKey": "MiniMax:MiniMax-M2", "inputUsdPer1M": 0.3, @@ -41081,6 +40097,12 @@ "inputUsdPer1M": 0.1, "outputUsdPer1M": 0.3 }, + { + "modelKey": "mistral:zai-glm-5-2", + "inputUsdPer1M": 1.4, + "outputUsdPer1M": 4.4, + "cacheReadUsdPer1M": 0.14 + }, { "modelKey": "moonshot:kimi-k2-0711-preview", "inputUsdPer1M": 0.6, @@ -41182,12 +40204,22 @@ "outputUsdPer1M": 0.28, "cacheReadUsdPer1M": 0.0028 }, + { + "modelKey": "nvidia:deepseek-ai/deepseek-v4-flash-0731", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "nvidia:deepseek-ai/deepseek-v4-pro", "inputUsdPer1M": 0.435, "outputUsdPer1M": 0.87, "cacheReadUsdPer1M": 0.003625 }, + { + "modelKey": "nvidia:deepseek-ai/deepseek-v4-pro-0813", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "nvidia:google/gemma-2-2b-it", "inputUsdPer1M": 0, @@ -41363,6 +40395,11 @@ "inputUsdPer1M": 0, "outputUsdPer1M": 0 }, + { + "modelKey": "nvidia:moonshotai/kimi-k3", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "nvidia:nvidia/active-speaker-detection", "inputUsdPer1M": 0, @@ -42186,6 +41223,12 @@ "inputUsdPer1M": 0, "outputUsdPer1M": 0 }, + { + "modelKey": "opencode:ling-3.0-flash-fin-free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0, + "cacheReadUsdPer1M": 0 + }, { "modelKey": "opencode:ling-3.0-flash-free", "inputUsdPer1M": 0, @@ -42375,22 +41418,22 @@ }, { "modelKey": "openrouter:~deepseek/deepseek-v4-flash-latest", - "inputUsdPer1M": 0.065, - "outputUsdPer1M": 0.18, - "cacheReadUsdPer1M": 0.02 + "inputUsdPer1M": 0.05, + "outputUsdPer1M": 0.16, + "cacheReadUsdPer1M": 0.013 }, { "modelKey": "openrouter:~google/gemini-flash-latest", - "inputUsdPer1M": 0.375, - "outputUsdPer1M": 1.875, - "cacheReadUsdPer1M": 0.0375, - "cacheWriteUsdPer1M": 0.020833 + "inputUsdPer1M": 0.75, + "outputUsdPer1M": 3.75, + "cacheReadUsdPer1M": 0.075, + "cacheWriteUsdPer1M": 0.041667 }, { "modelKey": "openrouter:~moonshotai/kimi-latest", - "inputUsdPer1M": 2.6, - "outputUsdPer1M": 13, - "cacheReadUsdPer1M": 0.29 + "inputUsdPer1M": 2.55, + "outputUsdPer1M": 12.75, + "cacheReadUsdPer1M": 0.256 }, { "modelKey": "openrouter:~openai/gpt-mini-latest", @@ -42400,9 +41443,9 @@ }, { "modelKey": "openrouter:~z-ai/glm-latest", - "inputUsdPer1M": 1.4, - "outputUsdPer1M": 4.4, - "cacheReadUsdPer1M": 0.26 + "inputUsdPer1M": 1.17, + "outputUsdPer1M": 3.96, + "cacheReadUsdPer1M": 0.234 }, { "modelKey": "openrouter:aion-labs/aion-2.0", @@ -42427,11 +41470,6 @@ "inputUsdPer1M": 0.8, "outputUsdPer1M": 1.6 }, - { - "modelKey": "openrouter:allenai/olmo-3-32b-think", - "inputUsdPer1M": 0.15, - "outputUsdPer1M": 0.5 - }, { "modelKey": "openrouter:amazon/nova-2-lite-v1", "inputUsdPer1M": 0.3, @@ -42460,7 +41498,7 @@ }, { "modelKey": "openrouter:anthracite-org/magnum-v4-72b", - "inputUsdPer1M": 3, + "inputUsdPer1M": 2.5, "outputUsdPer1M": 5 }, { @@ -42549,15 +41587,10 @@ }, { "modelKey": "openrouter:arcee-ai/trinity-large-thinking", - "inputUsdPer1M": 0.22, - "outputUsdPer1M": 0.85, + "inputUsdPer1M": 0.25, + "outputUsdPer1M": 0.8, "cacheReadUsdPer1M": 0.06 }, - { - "modelKey": "openrouter:arcee-ai/virtuoso-large", - "inputUsdPer1M": 0.75, - "outputUsdPer1M": 1.2 - }, { "modelKey": "openrouter:baidu/ernie-4.5-vl-424b-a47b", "inputUsdPer1M": 0.42, @@ -42604,11 +41637,6 @@ "inputUsdPer1M": 0, "outputUsdPer1M": 0 }, - { - "modelKey": "openrouter:deepcogito/cogito-v2.1-671b", - "inputUsdPer1M": 1.25, - "outputUsdPer1M": 1.25 - }, { "modelKey": "openrouter:deepseek/deepseek-chat", "inputUsdPer1M": 0.2574, @@ -42621,9 +41649,9 @@ }, { "modelKey": "openrouter:deepseek/deepseek-chat-v3.1", - "inputUsdPer1M": 0.25, - "outputUsdPer1M": 0.95, - "cacheReadUsdPer1M": 0.13 + "inputUsdPer1M": 0.55, + "outputUsdPer1M": 1.65, + "cacheReadUsdPer1M": 0.55 }, { "modelKey": "openrouter:deepseek/deepseek-r1", @@ -42660,21 +41688,21 @@ }, { "modelKey": "openrouter:deepseek/deepseek-v4-flash", - "inputUsdPer1M": 0.0826, - "outputUsdPer1M": 0.1652, - "cacheReadUsdPer1M": 0.01652 + "inputUsdPer1M": 0.08092, + "outputUsdPer1M": 0.16184, + "cacheReadUsdPer1M": 0.016184 }, { "modelKey": "openrouter:deepseek/deepseek-v4-flash-0731", - "inputUsdPer1M": 0.08, + "inputUsdPer1M": 0.065, "outputUsdPer1M": 0.18, "cacheReadUsdPer1M": 0.016 }, { "modelKey": "openrouter:deepseek/deepseek-v4-flash-vision-exp", - "inputUsdPer1M": 0.22, - "outputUsdPer1M": 0.66, - "cacheReadUsdPer1M": 0.007 + "inputUsdPer1M": 0.44, + "outputUsdPer1M": 1.32, + "cacheReadUsdPer1M": 0.014 }, { "modelKey": "openrouter:deepseek/deepseek-v4-pro", @@ -42684,9 +41712,9 @@ }, { "modelKey": "openrouter:deepseek/deepseek-v4-pro-0813", - "inputUsdPer1M": 1.188, - "outputUsdPer1M": 3.564, - "cacheReadUsdPer1M": 0.0396 + "inputUsdPer1M": 1.1154, + "outputUsdPer1M": 3.3462, + "cacheReadUsdPer1M": 0.03718 }, { "modelKey": "openrouter:dots-studio/dots-3-note-preview:free", @@ -42787,10 +41815,10 @@ }, { "modelKey": "openrouter:google/gemini-3.7-flash", - "inputUsdPer1M": 0.375, - "outputUsdPer1M": 1.875, - "cacheReadUsdPer1M": 0.0375, - "cacheWriteUsdPer1M": 0.020833 + "inputUsdPer1M": 0.75, + "outputUsdPer1M": 3.75, + "cacheReadUsdPer1M": 0.075, + "cacheWriteUsdPer1M": 0.041667 }, { "modelKey": "openrouter:google/gemma-2-27b-it", @@ -42813,11 +41841,6 @@ "inputUsdPer1M": 0.05, "outputUsdPer1M": 0.1 }, - { - "modelKey": "openrouter:google/gemma-3n-e4b-it", - "inputUsdPer1M": 0.06, - "outputUsdPer1M": 0.12 - }, { "modelKey": "openrouter:google/gemma-4-26b-a4b-it", "inputUsdPer1M": 0.07, @@ -42830,9 +41853,9 @@ }, { "modelKey": "openrouter:google/gemma-4-31b-it", - "inputUsdPer1M": 0.1, + "inputUsdPer1M": 0.09, "outputUsdPer1M": 0.34, - "cacheReadUsdPer1M": 0.1 + "cacheReadUsdPer1M": 0.05 }, { "modelKey": "openrouter:google/gemma-4-31b-it:free", @@ -42865,24 +41888,18 @@ "outputUsdPer1M": 0.1, "cacheReadUsdPer1M": 0.05 }, + { + "modelKey": "openrouter:ibm-granite/granite-4.2-8b", + "inputUsdPer1M": 0.1, + "outputUsdPer1M": 0.15, + "cacheReadUsdPer1M": 0.05 + }, { "modelKey": "openrouter:inception/mercury-2", "inputUsdPer1M": 0.25, "outputUsdPer1M": 0.75, "cacheReadUsdPer1M": 0.025 }, - { - "modelKey": "openrouter:inclusionai/ling-2.6-1t", - "inputUsdPer1M": 0.075, - "outputUsdPer1M": 0.625, - "cacheReadUsdPer1M": 0.015 - }, - { - "modelKey": "openrouter:inclusionai/ling-2.6-flash", - "inputUsdPer1M": 0.01, - "outputUsdPer1M": 0.03, - "cacheReadUsdPer1M": 0.002 - }, { "modelKey": "openrouter:inclusionai/ling-3.0-flash", "inputUsdPer1M": 0.021, @@ -42890,16 +41907,9 @@ "cacheReadUsdPer1M": 0.0042 }, { - "modelKey": "openrouter:inclusionai/ring-2.6-1t", - "inputUsdPer1M": 0.075, - "outputUsdPer1M": 0.625, - "cacheReadUsdPer1M": 0.015 - }, - { - "modelKey": "openrouter:kwaipilot/kat-coder-air-v2.5", - "inputUsdPer1M": 0.15, - "outputUsdPer1M": 0.6, - "cacheReadUsdPer1M": 0.03 + "modelKey": "openrouter:inclusionai/ling-3.0-flash-fin:free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 }, { "modelKey": "openrouter:kwaipilot/kat-coder-pro-v2", @@ -42920,7 +41930,7 @@ }, { "modelKey": "openrouter:mancer/weaver", - "inputUsdPer1M": 0.5, + "inputUsdPer1M": 0.4, "outputUsdPer1M": 0.75 }, { @@ -42952,13 +41962,14 @@ }, { "modelKey": "openrouter:meta-llama/llama-3.3-70b-instruct", - "inputUsdPer1M": 0.1, - "outputUsdPer1M": 0.32 + "inputUsdPer1M": 0.71, + "outputUsdPer1M": 0.71, + "cacheReadUsdPer1M": 0.71 }, { "modelKey": "openrouter:meta-llama/llama-4-maverick", "inputUsdPer1M": 0.2, - "outputUsdPer1M": 0.8 + "outputUsdPer1M": 0.696 }, { "modelKey": "openrouter:meta-llama/llama-4-scout", @@ -42973,7 +41984,7 @@ { "modelKey": "openrouter:meta/muse-glimmer-30b", "inputUsdPer1M": 0.3, - "outputUsdPer1M": 1.1, + "outputUsdPer1M": 1.2, "cacheReadUsdPer1M": 0.04 }, { @@ -42988,6 +41999,12 @@ "outputUsdPer1M": 4.25, "cacheReadUsdPer1M": 0.15 }, + { + "modelKey": "openrouter:meta/muse-spark-1.2-contributor", + "inputUsdPer1M": 0.1, + "outputUsdPer1M": 0.2, + "cacheReadUsdPer1M": 0.002 + }, { "modelKey": "openrouter:microsoft/phi-4", "inputUsdPer1M": 0.07, @@ -43028,8 +42045,8 @@ { "modelKey": "openrouter:minimax/minimax-m2.5", "inputUsdPer1M": 0.27, - "outputUsdPer1M": 0.95, - "cacheReadUsdPer1M": 0.03 + "outputUsdPer1M": 1.08, + "cacheReadUsdPer1M": 0.027 }, { "modelKey": "openrouter:minimax/minimax-m2.7", @@ -43037,18 +42054,34 @@ "outputUsdPer1M": 1.2, "cacheReadUsdPer1M": 0.06 }, + { + "modelKey": "openrouter:minimax/minimax-m2.7:free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "openrouter:minimax/minimax-m3", "inputUsdPer1M": 0.3, "outputUsdPer1M": 1.2, "cacheReadUsdPer1M": 0.06 }, + { + "modelKey": "openrouter:minimax/minimax-m3:free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "openrouter:mistralai/codestral-2508", "inputUsdPer1M": 0.3, "outputUsdPer1M": 0.9, "cacheReadUsdPer1M": 0.03 }, + { + "modelKey": "openrouter:mistralai/devstral-2512", + "inputUsdPer1M": 0.4, + "outputUsdPer1M": 2, + "cacheReadUsdPer1M": 0.04 + }, { "modelKey": "openrouter:mistralai/ministral-14b-2512", "inputUsdPer1M": 0.2, @@ -43061,11 +42094,6 @@ "outputUsdPer1M": 0.1, "cacheReadUsdPer1M": 0.01 }, - { - "modelKey": "openrouter:mistralai/ministral-8b", - "inputUsdPer1M": 0.11, - "outputUsdPer1M": 0.11 - }, { "modelKey": "openrouter:mistralai/ministral-8b-2512", "inputUsdPer1M": 0.15, @@ -43136,8 +42164,8 @@ }, { "modelKey": "openrouter:mistralai/mistral-small-3.2-24b-instruct", - "inputUsdPer1M": 0.09375, - "outputUsdPer1M": 0.25 + "inputUsdPer1M": 0.075, + "outputUsdPer1M": 0.2 }, { "modelKey": "openrouter:mistralai/mixtral-8x22b-instruct", @@ -43181,9 +42209,9 @@ }, { "modelKey": "openrouter:moonshotai/kimi-k2.7-code", - "inputUsdPer1M": 0.67, + "inputUsdPer1M": 0.66, "outputUsdPer1M": 3.4, - "cacheReadUsdPer1M": 0.17 + "cacheReadUsdPer1M": 0.18 }, { "modelKey": "openrouter:moonshotai/kimi-k3", @@ -43237,12 +42265,7 @@ "modelKey": "openrouter:nvidia/nemotron-3-nano-30b-a3b", "inputUsdPer1M": 0.05, "outputUsdPer1M": 0.2, - "cacheReadUsdPer1M": 0.03 - }, - { - "modelKey": "openrouter:nvidia/nemotron-3-nano-30b-a3b:free", - "inputUsdPer1M": 0, - "outputUsdPer1M": 0 + "cacheReadUsdPer1M": 0.025 }, { "modelKey": "openrouter:nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free", @@ -43261,9 +42284,9 @@ }, { "modelKey": "openrouter:nvidia/nemotron-3-ultra-550b-a55b", - "inputUsdPer1M": 0.6, - "outputUsdPer1M": 3.6, - "cacheReadUsdPer1M": 0.2 + "inputUsdPer1M": 0.5, + "outputUsdPer1M": 2.2, + "cacheReadUsdPer1M": 0.1 }, { "modelKey": "openrouter:nvidia/nemotron-3-ultra-550b-a55b:free", @@ -43286,16 +42309,6 @@ "inputUsdPer1M": 0, "outputUsdPer1M": 0 }, - { - "modelKey": "openrouter:nvidia/nemotron-nano-12b-v2-vl:free", - "inputUsdPer1M": 0, - "outputUsdPer1M": 0 - }, - { - "modelKey": "openrouter:nvidia/nemotron-nano-9b-v2:free", - "inputUsdPer1M": 0, - "outputUsdPer1M": 0 - }, { "modelKey": "openrouter:openai/gpt-3.5-turbo", "inputUsdPer1M": 0.5, @@ -43508,9 +42521,8 @@ }, { "modelKey": "openrouter:openai/gpt-oss-120b", - "inputUsdPer1M": 0.03, - "outputUsdPer1M": 0.17, - "cacheReadUsdPer1M": 0.03 + "inputUsdPer1M": 0.037, + "outputUsdPer1M": 0.17 }, { "modelKey": "openrouter:openai/gpt-oss-20b", @@ -43518,11 +42530,6 @@ "outputUsdPer1M": 0.13, "cacheReadUsdPer1M": 0.03 }, - { - "modelKey": "openrouter:openai/gpt-oss-20b:free", - "inputUsdPer1M": 0, - "outputUsdPer1M": 0 - }, { "modelKey": "openrouter:openai/gpt-oss-safeguard-20b", "inputUsdPer1M": 0.075, @@ -43649,9 +42656,8 @@ }, { "modelKey": "openrouter:qwen/qwen2.5-vl-72b-instruct", - "inputUsdPer1M": 0.8, - "outputUsdPer1M": 1, - "cacheReadUsdPer1M": 0.4 + "inputUsdPer1M": 0.25, + "outputUsdPer1M": 0.75 }, { "modelKey": "openrouter:qwen/qwen3-14b", @@ -43665,8 +42671,9 @@ }, { "modelKey": "openrouter:qwen/qwen3-235b-a22b-2507", - "inputUsdPer1M": 0.09, - "outputUsdPer1M": 0.55 + "inputUsdPer1M": 0.0875, + "outputUsdPer1M": 0.35, + "cacheReadUsdPer1M": 0.0175 }, { "modelKey": "openrouter:qwen/qwen3-235b-a22b-thinking-2507", @@ -43675,8 +42682,8 @@ }, { "modelKey": "openrouter:qwen/qwen3-30b-a3b", - "inputUsdPer1M": 0.13, - "outputUsdPer1M": 0.52 + "inputUsdPer1M": 0.12, + "outputUsdPer1M": 0.5 }, { "modelKey": "openrouter:qwen/qwen3-30b-a3b-instruct-2507", @@ -43717,8 +42724,9 @@ }, { "modelKey": "openrouter:qwen/qwen3-next-80b-a3b-instruct", - "inputUsdPer1M": 0.09, - "outputUsdPer1M": 1.1 + "inputUsdPer1M": 0.1, + "outputUsdPer1M": 1.1, + "cacheReadUsdPer1M": 0.07 }, { "modelKey": "openrouter:qwen/qwen3-next-80b-a3b-thinking", @@ -43738,8 +42746,8 @@ }, { "modelKey": "openrouter:qwen/qwen3-vl-30b-a3b-instruct", - "inputUsdPer1M": 0.13, - "outputUsdPer1M": 0.52 + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.6 }, { "modelKey": "openrouter:qwen/qwen3-vl-30b-a3b-thinking", @@ -43763,8 +42771,8 @@ }, { "modelKey": "openrouter:qwen/qwen3.5-122b-a10b", - "inputUsdPer1M": 0.26, - "outputUsdPer1M": 2.08 + "inputUsdPer1M": 0.29, + "outputUsdPer1M": 2.4 }, { "modelKey": "openrouter:qwen/qwen3.5-27b", @@ -43800,8 +42808,8 @@ }, { "modelKey": "openrouter:qwen/qwen3.6-35b-a3b", - "inputUsdPer1M": 0.14, - "outputUsdPer1M": 1, + "inputUsdPer1M": 0.1, + "outputUsdPer1M": 0.9, "cacheReadUsdPer1M": 0.05 }, { @@ -43819,9 +42827,17 @@ }, { "modelKey": "openrouter:qwen/qwen3.8-27b", - "inputUsdPer1M": 0.45, - "outputUsdPer1M": 3.2, - "cacheReadUsdPer1M": 0.05 + "inputUsdPer1M": 0.425, + "outputUsdPer1M": 2.55, + "cacheReadUsdPer1M": 0.085, + "cacheWriteUsdPer1M": 0.53125 + }, + { + "modelKey": "openrouter:qwen/qwen3.8-flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.47, + "cacheReadUsdPer1M": 0.016, + "cacheWriteUsdPer1M": 0.2 }, { "modelKey": "openrouter:qwen/qwen3.8-max", @@ -43871,11 +42887,6 @@ "inputUsdPer1M": 0.65, "outputUsdPer1M": 0.75 }, - { - "modelKey": "openrouter:stealth/ox-alpha", - "inputUsdPer1M": 0, - "outputUsdPer1M": 0 - }, { "modelKey": "openrouter:stepfun/step-3.5-flash", "inputUsdPer1M": 0.1, @@ -43902,6 +42913,11 @@ "inputUsdPer1M": 0.074, "outputUsdPer1M": 0.295 }, + { + "modelKey": "openrouter:tencent/hy-mt2-7b", + "inputUsdPer1M": 0.074, + "outputUsdPer1M": 0.295 + }, { "modelKey": "openrouter:tencent/hy3", "inputUsdPer1M": 0.132, @@ -43914,17 +42930,18 @@ "outputUsdPer1M": 0.6, "cacheReadUsdPer1M": 0.06 }, + { + "modelKey": "openrouter:tencent/hy4-preview", + "inputUsdPer1M": 0.834, + "outputUsdPer1M": 2.501, + "cacheReadUsdPer1M": 0.042 + }, { "modelKey": "openrouter:thedrummer/cydonia-24b-v4.1", "inputUsdPer1M": 0.3, "outputUsdPer1M": 0.5, "cacheReadUsdPer1M": 0.15 }, - { - "modelKey": "openrouter:thedrummer/rocinante-12b", - "inputUsdPer1M": 0.25, - "outputUsdPer1M": 0.5 - }, { "modelKey": "openrouter:thedrummer/skyfall-36b-v2", "inputUsdPer1M": 0.55, @@ -43938,9 +42955,9 @@ }, { "modelKey": "openrouter:thinkingmachines/inkling", - "inputUsdPer1M": 0.95, + "inputUsdPer1M": 1, "outputUsdPer1M": 4.05, - "cacheReadUsdPer1M": 0.16 + "cacheReadUsdPer1M": 0.17 }, { "modelKey": "openrouter:thinkingmachines/inkling-small", @@ -43948,6 +42965,16 @@ "outputUsdPer1M": 1.2, "cacheReadUsdPer1M": 0.1 }, + { + "modelKey": "openrouter:thinkingmachines/inkling-small:free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, + { + "modelKey": "openrouter:thinkingmachines/inkling:free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "openrouter:undi95/remm-slerp-l2-13b", "inputUsdPer1M": 0.45, @@ -44002,9 +43029,9 @@ }, { "modelKey": "openrouter:z-ai/glm-4.6", - "inputUsdPer1M": 0.5, - "outputUsdPer1M": 2, - "cacheReadUsdPer1M": 0.1 + "inputUsdPer1M": 0.43, + "outputUsdPer1M": 1.75, + "cacheReadUsdPer1M": 0.08 }, { "modelKey": "openrouter:z-ai/glm-4.6v", @@ -44044,9 +43071,9 @@ }, { "modelKey": "openrouter:z-ai/glm-5.2", - "inputUsdPer1M": 0.966, - "outputUsdPer1M": 3.036, - "cacheReadUsdPer1M": 0.1932 + "inputUsdPer1M": 1.19, + "outputUsdPer1M": 3.74, + "cacheReadUsdPer1M": 0.221 }, { "modelKey": "openrouter:z-ai/glm-5.2:free", @@ -44059,6 +43086,12 @@ "outputUsdPer1M": 4.4, "cacheReadUsdPer1M": 0.26 }, + { + "modelKey": "openrouter:z-ai/glm-5.3-flash", + "inputUsdPer1M": 0.075, + "outputUsdPer1M": 0.25, + "cacheReadUsdPer1M": 0.015 + }, { "modelKey": "openrouter:z-ai/glm-5v-turbo", "inputUsdPer1M": 1.2, @@ -44578,6 +43611,18 @@ "outputUsdPer1M": 4.4, "cacheReadUsdPer1M": 0.26 }, + { + "modelKey": "togetherai:zai-org/GLM-5.3", + "inputUsdPer1M": 1.4, + "outputUsdPer1M": 4.4, + "cacheReadUsdPer1M": 0.26 + }, + { + "modelKey": "togetherai:zai-org/GLM-5.3-Flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.5, + "cacheReadUsdPer1M": 0.03 + }, { "modelKey": "tencent-tokenhub:hy3", "inputUsdPer1M": 0, @@ -44592,6 +43637,12 @@ "cacheReadUsdPer1M": 0, "cacheWriteUsdPer1M": 0 }, + { + "modelKey": "tencent-tokenhub:hy4-preview", + "inputUsdPer1M": 0.834, + "outputUsdPer1M": 2.501, + "cacheReadUsdPer1M": 0.042 + }, { "modelKey": "vercel:alibaba/qwen-3-14b", "inputUsdPer1M": 0.12, @@ -44740,13 +43791,27 @@ "modelKey": "vercel:alibaba/qwen3.8-2.4t-a95b", "inputUsdPer1M": 2, "outputUsdPer1M": 6, - "cacheReadUsdPer1M": 0.2 + "cacheReadUsdPer1M": 0.25 }, { "modelKey": "vercel:alibaba/qwen3.8-27b", - "inputUsdPer1M": 0.55, - "outputUsdPer1M": 3.3, - "cacheReadUsdPer1M": 0.11 + "inputUsdPer1M": 0.5, + "outputUsdPer1M": 3, + "cacheReadUsdPer1M": 0.1, + "cacheWriteUsdPer1M": 0.625 + }, + { + "modelKey": "vercel:alibaba/qwen3.8-flash", + "inputUsdPer1M": 0.16, + "outputUsdPer1M": 0.47, + "cacheReadUsdPer1M": 0.016, + "cacheWriteUsdPer1M": 0.2 + }, + { + "modelKey": "vercel:alibaba/qwen3.8-flash-next", + "inputUsdPer1M": 0.12, + "outputUsdPer1M": 0.4, + "cacheReadUsdPer1M": 0.01 }, { "modelKey": "vercel:alibaba/qwen3.8-max", @@ -44882,11 +43947,6 @@ "inputUsdPer1M": 0.25, "outputUsdPer1M": 0.8999999999999999 }, - { - "modelKey": "vercel:arcee-ai/trinity-mini", - "inputUsdPer1M": 0.045, - "outputUsdPer1M": 0.15 - }, { "modelKey": "vercel:bytedance/seed-1.6", "inputUsdPer1M": 0.25, @@ -44946,21 +44006,27 @@ }, { "modelKey": "vercel:deepseek/deepseek-v4-flash-0731", - "inputUsdPer1M": 0.13, - "outputUsdPer1M": 0.26, - "cacheReadUsdPer1M": 0.028 + "inputUsdPer1M": 0.076, + "outputUsdPer1M": 0.153, + "cacheReadUsdPer1M": 0.014 + }, + { + "modelKey": "vercel:deepseek/deepseek-v4-flash-vision-exp", + "inputUsdPer1M": 0.22, + "outputUsdPer1M": 0.66, + "cacheReadUsdPer1M": 0.007 }, { "modelKey": "vercel:deepseek/deepseek-v4-pro", - "inputUsdPer1M": 1.74, - "outputUsdPer1M": 3.48, - "cacheReadUsdPer1M": 0.14 + "inputUsdPer1M": 0.66, + "outputUsdPer1M": 1.98, + "cacheReadUsdPer1M": 0.022 }, { "modelKey": "vercel:deepseek/deepseek-v4-pro-0813", - "inputUsdPer1M": 1.32, - "outputUsdPer1M": 3.96, - "cacheReadUsdPer1M": 0.132 + "inputUsdPer1M": 0.66, + "outputUsdPer1M": 1.98, + "cacheReadUsdPer1M": 0.066 }, { "modelKey": "vercel:google/gemini-2.5-flash", @@ -45034,6 +44100,11 @@ "outputUsdPer1M": 2.5, "cacheReadUsdPer1M": 0.03 }, + { + "modelKey": "vercel:google/gemini-3.5-transcribe", + "inputUsdPer1M": 2, + "outputUsdPer1M": 12 + }, { "modelKey": "vercel:google/gemini-3.6-flash", "inputUsdPer1M": 0.75, @@ -45079,6 +44150,16 @@ "outputUsdPer1M": 0.18, "cacheReadUsdPer1M": 0.012 }, + { + "modelKey": "vercel:inclusionai/ling-3.0-flash-fin", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, + { + "modelKey": "vercel:inclusionai/ling-3.0-flash-fin-free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "vercel:interfaze/interfaze-beta", "inputUsdPer1M": 1.5, @@ -45199,6 +44280,11 @@ "cacheReadUsdPer1M": 0.06, "cacheWriteUsdPer1M": 0.375 }, + { + "modelKey": "vercel:minimax/minimax-m2.7-free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0 + }, { "modelKey": "vercel:minimax/minimax-m2.7-highspeed", "inputUsdPer1M": 0.6, @@ -45212,6 +44298,12 @@ "outputUsdPer1M": 1.2, "cacheReadUsdPer1M": 0.06 }, + { + "modelKey": "vercel:minimax/minimax-m3-free", + "inputUsdPer1M": 0, + "outputUsdPer1M": 0, + "cacheReadUsdPer1M": 0 + }, { "modelKey": "vercel:mistral/codestral", "inputUsdPer1M": 0.3, @@ -45227,16 +44319,6 @@ "inputUsdPer1M": 0.1, "outputUsdPer1M": 0.3 }, - { - "modelKey": "vercel:mistral/magistral-medium", - "inputUsdPer1M": 2, - "outputUsdPer1M": 5 - }, - { - "modelKey": "vercel:mistral/magistral-small", - "inputUsdPer1M": 0.5, - "outputUsdPer1M": 1.5 - }, { "modelKey": "vercel:mistral/ministral-14b", "inputUsdPer1M": 0.2, @@ -45309,7 +44391,7 @@ "modelKey": "vercel:moonshotai/kimi-k2.7-code", "inputUsdPer1M": 0.95, "outputUsdPer1M": 4, - "cacheReadUsdPer1M": 0.19 + "cacheReadUsdPer1M": 0.16 }, { "modelKey": "vercel:moonshotai/kimi-k2.7-code-highspeed", @@ -45441,11 +44523,6 @@ "outputUsdPer1M": 1, "cacheReadUsdPer1M": 0.125 }, - { - "modelKey": "vercel:openai/gpt-4o-mini-search-preview", - "inputUsdPer1M": 0.15, - "outputUsdPer1M": 0.6 - }, { "modelKey": "vercel:openai/gpt-4o-mini-transcribe", "inputUsdPer1M": 1.25, @@ -45630,17 +44707,17 @@ }, { "modelKey": "vercel:openai/gpt-5.6-sol", - "inputUsdPer1M": 2.5, - "outputUsdPer1M": 15, - "cacheReadUsdPer1M": 0.25, - "cacheWriteUsdPer1M": 3.125 + "inputUsdPer1M": 2, + "outputUsdPer1M": 10, + "cacheReadUsdPer1M": 0.2, + "cacheWriteUsdPer1M": 2.5 }, { "modelKey": "vercel:openai/gpt-5.6-sol-fast", - "inputUsdPer1M": 5, - "outputUsdPer1M": 30, - "cacheReadUsdPer1M": 0.5, - "cacheWriteUsdPer1M": 3.125 + "inputUsdPer1M": 4, + "outputUsdPer1M": 20, + "cacheReadUsdPer1M": 0.4, + "cacheWriteUsdPer1M": 2.5 }, { "modelKey": "vercel:openai/gpt-5.6-terra", @@ -45690,11 +44767,15 @@ "inputUsdPer1M": 0.05, "outputUsdPer1M": 0.2 }, + { + "modelKey": "vercel:openai/gpt-oss-safeguard-120b", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.6 + }, { "modelKey": "vercel:openai/gpt-oss-safeguard-20b", - "inputUsdPer1M": 0.075, - "outputUsdPer1M": 0.3, - "cacheReadUsdPer1M": 0.037 + "inputUsdPer1M": 0.07, + "outputUsdPer1M": 0.2 }, { "modelKey": "vercel:openai/gpt-realtime-1.5", @@ -45732,12 +44813,6 @@ "outputUsdPer1M": 8, "cacheReadUsdPer1M": 0.5 }, - { - "modelKey": "vercel:openai/o3-deep-research", - "inputUsdPer1M": 10, - "outputUsdPer1M": 40, - "cacheReadUsdPer1M": 2.5 - }, { "modelKey": "vercel:openai/o3-fast", "inputUsdPer1M": 3.5, @@ -45891,9 +44966,15 @@ }, { "modelKey": "vercel:tencent/hy3", - "inputUsdPer1M": 0.132, - "outputUsdPer1M": 0.528, - "cacheReadUsdPer1M": 0.033 + "inputUsdPer1M": 0.14, + "outputUsdPer1M": 0.58, + "cacheReadUsdPer1M": 0.035 + }, + { + "modelKey": "vercel:tencent/hy4-preview", + "inputUsdPer1M": 0.834, + "outputUsdPer1M": 2.501, + "cacheReadUsdPer1M": 0.042 }, { "modelKey": "vercel:thinkingmachines/inkling", @@ -45919,6 +45000,12 @@ "outputUsdPer1M": 0.87, "cacheReadUsdPer1M": 0.0036 }, + { + "modelKey": "vercel:xiaomi/mimo-v2.5-pro-ultraspeed", + "inputUsdPer1M": 1.305, + "outputUsdPer1M": 2.61, + "cacheReadUsdPer1M": 0.0108 + }, { "modelKey": "vercel:zai/glm-4.5", "inputUsdPer1M": 0.6, @@ -45993,7 +45080,13 @@ "modelKey": "vercel:zai/glm-5.3", "inputUsdPer1M": 1.4, "outputUsdPer1M": 4.4, - "cacheReadUsdPer1M": 0.26 + "cacheReadUsdPer1M": 0.14 + }, + { + "modelKey": "vercel:zai/glm-5.3-flash", + "inputUsdPer1M": 0.15, + "outputUsdPer1M": 0.5, + "cacheReadUsdPer1M": 0.03 }, { "modelKey": "vercel:zai/glm-5v-turbo", @@ -46124,6 +45217,20 @@ "cacheReadUsdPer1M": 0.26, "cacheWriteUsdPer1M": 0 }, + { + "modelKey": "zai:glm-5.3", + "inputUsdPer1M": 1.4, + "outputUsdPer1M": 4.4, + "cacheReadUsdPer1M": 0.26, + "cacheWriteUsdPer1M": 0 + }, + { + "modelKey": "zai:glm-5.3-flash", + "inputUsdPer1M": 0.075, + "outputUsdPer1M": 0.25, + "cacheReadUsdPer1M": 0.015, + "cacheWriteUsdPer1M": 0 + }, { "modelKey": "zai:glm-5v-turbo", "inputUsdPer1M": 1.2, @@ -47249,6 +46356,9 @@ "grok-4.5": { "npm": "@ai-sdk/openai" }, + "grok-4.6": { + "npm": "@ai-sdk/openai" + }, "minimax-m2.5": { "npm": "@ai-sdk/anthropic" }, @@ -47260,6 +46370,9 @@ }, "muse-spark-1.2-contributor": { "npm": "@ai-sdk/openai" + }, + "qwen3.8-flash": { + "npm": "@ai-sdk/anthropic" } }, "openrouter": {}, diff --git a/scripts/sync-model-metadata.mjs b/scripts/sync-model-metadata.mjs index 2869c04416..5203456772 100644 --- a/scripts/sync-model-metadata.mjs +++ b/scripts/sync-model-metadata.mjs @@ -25,6 +25,11 @@ import { dirname } from 'node:path'; import { pathToFileURL } from 'node:url'; const SOURCE_URL = 'https://models.dev/api.json'; +// This script runs before packages/core is built, so it cannot read +// ModelModality itself. The build keeps the two in step: a value that reaches +// the projection but not that type fails to compile. A value missing here +// drops the whole refresh, not just the model that declares it. +const MODALITIES = new Set(['text', 'image', 'audio', 'pdf', 'video']); const DEFAULT_SNAPSHOT = 'scripts/model-metadata/models-dev-api.snapshot.json'; const DEFAULT_OUTPUT = 'packages/core/src/model-metadata.generated.ts'; const DEFAULT_PRICING_OUTPUT = 'packages/runtime/src/telemetry/model-pricing.generated.ts'; @@ -113,11 +118,22 @@ export async function main(argv = process.argv) { (outputPath === DEFAULT_OUTPUT ? DEFAULT_PRICING_OUTPUT : undefined); const refresh = argv.includes('--refresh'); const check = argv.includes('--check'); + const drift = argv.includes('--drift'); const acceptUpstreamRemovals = argv.includes('--accept-upstream-removals'); - if (refreshInputPath && !refresh) throw new Error('--refresh-input requires --refresh'); + if (drift && (refresh || check)) { + throw new Error('--drift reports without writing and cannot combine with --refresh or --check'); + } + if (refreshInputPath && !refresh && !drift) { + throw new Error('--refresh-input requires --refresh or --drift'); + } if (acceptUpstreamRemovals && !refresh) { throw new Error('--accept-upstream-removals requires --refresh'); } + if (drift) { + const report = await collectDrift(await loadSnapshot(snapshotPath), refreshInputPath); + process.stdout.write(`${formatDrift(report)}\n`); + return report; + } const source = refresh ? await refreshSnapshot(snapshotPath, refreshInputPath, { acceptUpstreamRemovals }) @@ -176,30 +192,25 @@ export async function main(argv = process.argv) { await replaceFilesTransactionally(writes); } -function buildProjection(catalog) { +// `options.onReject` decides what an unprojectable provider or model costs. A +// refresh has none, so the first bad shape aborts the whole snapshot rather +// than silently committing a catalog with a hole in it. The drift report +// passes one, because there a bad shape is the finding it exists to print and +// must not stop it comparing everything else. +function buildProjection(catalog, options = {}) { + const onReject = options.onReject; const metadata = {}; const pricing = []; const providerFacts = {}; const providerOverrides = {}; for (const [providerType, sourceId] of Object.entries(PROVIDERS)) { const provider = catalog[sourceId]; - if (!provider) { - throw new Error(`models.dev provider ${sourceId} is missing`); - } - if ( - !provider.models || - typeof provider.models !== 'object' || - Array.isArray(provider.models) || - Object.keys(provider.models).length === 0 - ) { - throw new Error(`models.dev provider ${sourceId} has no non-empty models object`); - } - if ( - typeof provider.id !== 'string' || - typeof provider.name !== 'string' || - typeof provider.doc !== 'string' - ) { - throw new Error(`models.dev provider ${sourceId} has an unsupported shape`); + try { + assertProviderShape(sourceId, provider); + } catch (error) { + if (!onReject) throw error; + onReject('provider', providerType, error); + continue; } providerFacts[providerType] = { id: provider.id, @@ -207,43 +218,78 @@ function buildProjection(catalog) { ...(typeof provider.api === 'string' ? { api: provider.api } : {}), doc: provider.doc, }; - metadata[providerType] = Object.fromEntries( - Object.entries(provider.models) - .sort(([left], [right]) => left.localeCompare(right)) - .map(([id, model]) => [id, toMetadata(sourceId, id, provider, model)]), - ); - providerOverrides[providerType] = Object.fromEntries( - Object.entries(provider.models) - .sort(([left], [right]) => left.localeCompare(right)) - .filter(([, model]) => model.provider !== undefined) - .map(([id, model]) => [id, toModelProviderOverride(sourceId, id, model.provider)]), + metadata[providerType] = {}; + providerOverrides[providerType] = {}; + const priced = !PRICING_EXCLUDED_PROVIDER_TYPES.has(providerType); + const models = Object.entries(provider.models).sort(([left], [right]) => + left.localeCompare(right), ); - if (!PRICING_EXCLUDED_PROVIDER_TYPES.has(providerType)) { - pricing.push( - ...Object.entries(provider.models) - .sort(([left], [right]) => left.localeCompare(right)) - .map(([id, model]) => toPricing(providerType, id, model)) - .filter((pricing) => pricing !== undefined), - ); + for (const [id, model] of models) { + let projected; + try { + projected = { + metadata: toMetadata(sourceId, id, provider, model), + override: + model.provider === undefined + ? undefined + : toModelProviderOverride(sourceId, id, model.provider), + pricing: priced ? toPricing(providerType, id, model) : undefined, + }; + } catch (error) { + if (!onReject) throw error; + onReject('model', `${providerType}/${id}`, error); + continue; + } + metadata[providerType][id] = projected.metadata; + if (projected.override !== undefined) + providerOverrides[providerType][id] = projected.override; + if (projected.pricing !== undefined) pricing.push(projected.pricing); } } return { metadata, pricing, providerFacts, providerOverrides }; } -async function refreshSnapshot(snapshotPath, refreshInputPath, options = {}) { - let sourceText; - let sourceEtag = null; - let retrievedAt = new Date().toISOString(); +function assertProviderShape(sourceId, provider) { + if (!provider) { + throw new Error(`models.dev provider ${sourceId} is missing`); + } + if ( + !provider.models || + typeof provider.models !== 'object' || + Array.isArray(provider.models) || + Object.keys(provider.models).length === 0 + ) { + throw new Error(`models.dev provider ${sourceId} has no non-empty models object`); + } + if ( + typeof provider.id !== 'string' || + typeof provider.name !== 'string' || + typeof provider.doc !== 'string' + ) { + throw new Error(`models.dev provider ${sourceId} has an unsupported shape`); + } +} + +async function readUpstream(refreshInputPath) { if (refreshInputPath) { - sourceText = await readFile(refreshInputPath, 'utf8'); - } else { - const response = await fetch(SOURCE_URL, { signal: AbortSignal.timeout(10_000) }); - if (!response.ok) throw new Error(`models.dev returned HTTP ${response.status}`); - sourceText = await response.text(); - sourceEtag = response.headers.get('etag'); - retrievedAt = new Date(response.headers.get('date') ?? Date.now()).toISOString(); + return { + text: await readFile(refreshInputPath, 'utf8'), + etag: null, + retrievedAt: new Date().toISOString(), + }; } + const response = await fetch(SOURCE_URL, { signal: AbortSignal.timeout(10_000) }); + if (!response.ok) throw new Error(`models.dev returned HTTP ${response.status}`); + return { + text: await response.text(), + etag: response.headers.get('etag'), + retrievedAt: new Date(response.headers.get('date') ?? Date.now()).toISOString(), + }; +} + +async function refreshSnapshot(snapshotPath, refreshInputPath, options = {}) { + const { text: sourceText, etag: sourceEtag, retrievedAt } = await readUpstream(refreshInputPath); const projection = buildProjection(selectCatalog(JSON.parse(sourceText))); if (!options.acceptUpstreamRemovals) { const previous = await loadSnapshotIfPresent(snapshotPath); @@ -335,6 +381,134 @@ function projectionPath(path) { return `/${path.map((segment) => String(segment).replaceAll('~', '~0').replaceAll('/', '~1')).join('/')}`; } +// `--check` only proves the generated modules match the committed snapshot. +// Nothing compared that snapshot against models.dev, which is how it stayed +// weeks behind upstream without anything reporting it. This walks the two one +// model at a time, so a shape the projector rejects becomes its own finding +// instead of aborting the whole comparison the way a refresh does. +const DRIFT_LIST_LIMIT = 20; + +// Every section of the projection, flattened to one value per entity. The +// report reads sections through this table instead of naming them itself, +// which is how it went out comparing only two of the four. A section added to +// buildProjection is compared here without touching the comparison. +const PROJECTION_SECTIONS = { + metadata: entitiesByProviderAndModel, + // modelKey is `${providerType}:${id}`, so replacing the first colon yields + // the label every other section already uses. + pricing: (section) => new Map(section.map((entry) => [entry.modelKey.replace(':', '/'), entry])), + providerFacts: (section) => new Map(Object.entries(section)), + providerOverrides: entitiesByProviderAndModel, +}; + +async function collectDrift(snapshot, refreshInputPath) { + const rejectedProviders = []; + const rejectedModels = []; + const rejected = new Set(); + // The raw catalog, not selectCatalog's: a provider that vanished upstream is + // the report's most important finding, and selectCatalog throws on it. + const upstream = buildProjection(JSON.parse((await readUpstream(refreshInputPath)).text), { + onReject: (kind, label, error) => { + rejected.add(label); + (kind === 'provider' ? rejectedProviders : rejectedModels).push(`${label}: ${error.message}`); + }, + }); + const previous = projectionEntities(snapshot.projection); + const next = projectionEntities(upstream); + const report = { rejectedProviders, rejectedModels, added: [], removed: [], changed: [] }; + for (const label of [...new Set([...previous.keys(), ...next.keys()])].sort()) { + const before = previous.get(label); + const after = next.get(label); + if (before === undefined) { + report.added.push(label); + continue; + } + if (after === undefined) { + // A shape the projector rejected already has its own finding above. It + // is not upstream saying the entity is gone. + if (!isRejected(label, rejected)) report.removed.push(label); + continue; + } + const fields = driftedFields(before, after); + if (fields.length > 0) report.changed.push(`${label}: ${fields.join(', ')}`); + } + const drifted = Object.values(report).some((entries) => entries.length > 0); + return { ...report, drifted }; +} + +function entitiesByProviderAndModel(section) { + const entities = new Map(); + for (const [providerType, models] of Object.entries(section)) { + for (const [id, value] of Object.entries(models)) entities.set(`${providerType}/${id}`, value); + } + return entities; +} + +function projectionEntities(projection) { + const entities = new Map(); + for (const [section, flatten] of Object.entries(PROJECTION_SECTIONS)) { + for (const [label, value] of flatten(projection[section])) { + const entity = entities.get(label); + if (entity) entity[section] = value; + else entities.set(label, { [section]: value }); + } + } + return entities; +} + +function isRejected(label, rejected) { + if (rejected.has(label)) return true; + const slash = label.indexOf('/'); + return slash !== -1 && rejected.has(label.slice(0, slash)); +} + +function driftedFields(previous, next) { + const fields = []; + for (const section of Object.keys(PROJECTION_SECTIONS)) { + const before = previous[section]; + const after = next[section]; + if (sameValue(before, after)) continue; + if (isPlainObject(before) && isPlainObject(after)) { + const keys = [...new Set([...Object.keys(before), ...Object.keys(after)])].sort(); + for (const key of keys) { + if (!sameValue(before[key], after[key])) fields.push(`${section}.${key}`); + } + continue; + } + fields.push(section); + } + return fields; +} + +function isPlainObject(value) { + return typeof value === 'object' && value !== null && !Array.isArray(value); +} + +// Both sides are projector output, so their keys are already in one order. +function sameValue(left, right) { + return JSON.stringify(left) === JSON.stringify(right); +} + +function formatDrift(report) { + const lines = []; + for (const [label, entries] of [ + ['providers the projector rejects', report.rejectedProviders], + ['models the projector rejects', report.rejectedModels], + ['entries upstream has and the snapshot does not', report.added], + ['entries the snapshot has and upstream does not', report.removed], + ['entries whose projection changed', report.changed], + ]) { + if (entries.length === 0) continue; + lines.push(`${label}: ${entries.length}`); + for (const entry of entries.slice(0, DRIFT_LIST_LIMIT)) lines.push(` ${entry}`); + if (entries.length > DRIFT_LIST_LIMIT) { + lines.push(` ... and ${entries.length - DRIFT_LIST_LIMIT} more`); + } + } + if (!report.drifted) return `${SOURCE_URL} matches the committed snapshot.`; + return [`${SOURCE_URL} has drifted from the committed snapshot.`, ...lines].join('\n'); +} + async function replaceFilesTransactionally(writes) { if (new Set(writes.map((write) => write.path)).size !== writes.length) { throw new Error('model metadata outputs must use distinct paths'); @@ -540,12 +714,8 @@ export function toMetadata(providerId, modelId, provider, model) { throw new Error(`models.dev model ${providerId}/${modelId} has an unsupported shape`); } if ( - model.modalities?.input.some( - (value) => value !== 'text' && value !== 'image' && value !== 'audio' && value !== 'pdf', - ) || - model.modalities?.output.some( - (value) => value !== 'text' && value !== 'image' && value !== 'audio', - ) + model.modalities?.input.some((value) => !MODALITIES.has(value)) || + model.modalities?.output.some((value) => !MODALITIES.has(value)) ) { throw new Error(`models.dev model ${providerId}/${modelId} has unsupported modalities`); } @@ -585,7 +755,6 @@ export function toMetadata(providerId, modelId, provider, model) { displayName: model.name, ...(model.description !== undefined ? { description: model.description } : {}), lifecycle, - docsUrl: provider.doc, contextWindow: model.limit?.context, ...(model.limit?.input !== undefined ? { inputLimit: model.limit.input } : {}), maxOutputTokens: model.limit?.output, @@ -706,7 +875,10 @@ function buildPricingModule(pricing, source) { } if (process.argv[1] && import.meta.url === pathToFileURL(process.argv[1]).href) { - await main(); + // Drift is a finding, not a crash. Exit 2 so a caller can tell "upstream + // moved" from "this command failed"; both reported 1 before, which made the + // difference unreadable to the one job that has to act on it. + if ((await main())?.drifted) process.exitCode = 2; } function option(name, argv) { diff --git a/scripts/sync-model-metadata.test.mjs b/scripts/sync-model-metadata.test.mjs index 021b2aaf85..52c1e35cbf 100644 --- a/scripts/sync-model-metadata.test.mjs +++ b/scripts/sync-model-metadata.test.mjs @@ -280,7 +280,7 @@ test('refresh rejects unknown model modalities instead of dropping them', async const snapshot = join(root, 'snapshot.json'); const metadata = join(root, 'metadata.ts'); const catalog = fixtureCatalog(); - catalog.anthropic.models.model.modalities = { input: ['text', 'video'], output: ['text'] }; + catalog.anthropic.models.model.modalities = { input: ['text', 'hologram'], output: ['text'] }; await writeFile(input, JSON.stringify(catalog)); await assert.rejects( @@ -302,6 +302,182 @@ test('refresh rejects unknown model modalities instead of dropping them', async } }); +// The drift report goes to stdout for the scheduled job's summary; tests read +// the returned buckets instead of the printed text. +async function drift(argv) { + const written = []; + const original = process.stdout.write; + process.stdout.write = (chunk) => { + written.push(String(chunk)); + return true; + }; + try { + return { report: await main(['node', 'sync-model-metadata.mjs', '--drift', ...argv]) }; + } finally { + process.stdout.write = original; + assert.ok(written.join('').length > 0, 'drift must print a report'); + } +} + +test('drift finds nothing when the snapshot still matches upstream', async () => { + const root = await mkdtemp(join(tmpdir(), 'maka-model-drift-clean-')); + try { + const input = join(root, 'api.json'); + const snapshot = join(root, 'snapshot.json'); + await writeFile(input, JSON.stringify(fixtureCatalog())); + await main([ + 'node', + 'sync-model-metadata.mjs', + '--refresh', + '--refresh-input', + input, + '--snapshot', + snapshot, + '--output', + join(root, 'metadata.ts'), + ]); + + const { report } = await drift(['--snapshot', snapshot, '--refresh-input', input]); + assert.equal(report.drifted, false); + assert.deepEqual( + [ + report.added, + report.removed, + report.changed, + report.rejectedModels, + report.rejectedProviders, + ], + [[], [], [], [], []], + ); + } finally { + await rm(root, { recursive: true, force: true }); + } +}); + +test('drift separates a rejected shape from a real upstream difference', async () => { + const root = await mkdtemp(join(tmpdir(), 'maka-model-drift-')); + try { + const committed = join(root, 'api.json'); + const upstream = join(root, 'upstream.json'); + const snapshot = join(root, 'snapshot.json'); + const before = fixtureCatalog(); + before.anthropic.models.legacy = { ...before.anthropic.models.model, name: 'Legacy' }; + await writeFile(committed, JSON.stringify(before)); + await main([ + 'node', + 'sync-model-metadata.mjs', + '--refresh', + '--refresh-input', + committed, + '--snapshot', + snapshot, + '--output', + join(root, 'metadata.ts'), + ]); + + const after = JSON.parse(JSON.stringify(before)); + delete after.anthropic.models.legacy; + after.anthropic.models.added = { ...after.anthropic.models.model, name: 'Added' }; + after.anthropic.models.model.name = 'Renamed'; + after.anthropic.models.model.cost = { input: 9, output: 9 }; + after.groq.models.model.modalities = { input: ['text', 'hologram'], output: ['text'] }; + delete after.openai; + await writeFile(upstream, JSON.stringify(after)); + + const { report } = await drift(['--snapshot', snapshot, '--refresh-input', upstream]); + assert.equal(report.drifted, true); + assert.deepEqual(report.added, ['anthropic/added']); + assert.deepEqual(report.removed, ['anthropic/legacy']); + assert.deepEqual(report.changed, [ + 'anthropic/model: metadata.displayName, pricing.inputUsdPer1M, pricing.outputUsdPer1M', + ]); + assert.equal(report.rejectedProviders.length, 1); + assert.match(report.rejectedProviders[0], /^openai: .*provider openai is missing/u); + assert.equal(report.rejectedModels.length, 1); + assert.match(report.rejectedModels[0], /^groq\/model: .*unsupported modalities/u); + } finally { + await rm(root, { recursive: true, force: true }); + } +}); + +test('drift reports the projection sections that carry no model metadata', async () => { + const root = await mkdtemp(join(tmpdir(), 'maka-model-drift-sections-')); + try { + const committed = join(root, 'api.json'); + const upstream = join(root, 'upstream.json'); + const snapshot = join(root, 'snapshot.json'); + const before = fixtureCatalog(); + before.anthropic.models.model.provider = { npm: '@example/before' }; + await writeFile(committed, JSON.stringify(before)); + await main([ + 'node', + 'sync-model-metadata.mjs', + '--refresh', + '--refresh-input', + committed, + '--snapshot', + snapshot, + '--output', + join(root, 'metadata.ts'), + ]); + + // Neither a renamed provider nor a swapped npm package touches a model's + // metadata or its pricing, which is all the report used to compare. + const after = JSON.parse(JSON.stringify(before)); + after.anthropic.name = 'Anthropic Renamed'; + after.anthropic.models.model.provider = { npm: '@example/after' }; + await writeFile(upstream, JSON.stringify(after)); + + const { report } = await drift(['--snapshot', snapshot, '--refresh-input', upstream]); + assert.equal(report.drifted, true); + assert.deepEqual(report.changed, [ + 'anthropic: providerFacts.name', + 'anthropic/model: providerOverrides.npm', + ]); + } finally { + await rm(root, { recursive: true, force: true }); + } +}); + +test('drift refuses to combine with the modes that write', async () => { + for (const mode of ['--refresh', '--check']) { + await assert.rejects( + main(['node', 'sync-model-metadata.mjs', '--drift', mode]), + /--drift reports without writing/u, + ); + } +}); + +test('refresh carries the video and pdf modalities models.dev declares', async () => { + const root = await mkdtemp(join(tmpdir(), 'maka-model-snapshot-video-')); + try { + const input = join(root, 'api.json'); + const snapshot = join(root, 'snapshot.json'); + const metadata = join(root, 'metadata.ts'); + const catalog = fixtureCatalog(); + const modalities = { input: ['text', 'video'], output: ['text', 'pdf', 'video'] }; + catalog.anthropic.models.model.modalities = modalities; + await writeFile(input, JSON.stringify(catalog)); + + await main([ + 'node', + 'sync-model-metadata.mjs', + '--refresh', + '--refresh-input', + input, + '--snapshot', + snapshot, + '--output', + metadata, + ]); + + const written = JSON.parse(await readFile(snapshot, 'utf8')); + assert.deepEqual(written.projection.metadata.anthropic.model.modalities, modalities); + } finally { + await rm(root, { recursive: true, force: true }); + } +}); + test('refresh rejects a partial provider shrink until it is explicitly accepted', async () => { const root = await mkdtemp(join(tmpdir(), 'maka-model-snapshot-provider-shrink-')); try {