Repository navigation
feat(models): add deprecation notices for legacy Qwen and Gemini models - #3288
Merged
Merged
Conversation
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (2)
🚧 Files skipped from review as they are similar to previous changes (1)
WalkthroughAlibaba and Google model-provider catalog entries now include updated ChangesModel provider metadata
Estimated code review effort: 1 (Trivial) | ~5 minutes Possibly related PRs
Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
vicovaro
marked this pull request as ready for review
July 28, 2026 16:20
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@packages/models/src/models/alibaba.ts`:
- Line 2770: Remove the deactivatedAt property from the qwen3.6-plus model
definition, leaving its supported routing configuration unchanged; do not alter
the October 10, 2026 date for qwen3.6-max-preview.
- Line 1357: Remove the deactivatedAt metadata from both Novita mappings for
qwen/qwen3-max and qwen/qwen3-vl-235b-a22b-instruct at
packages/models/src/models/alibaba.ts lines 1357-1357 and 2433-2433; retain
retirement metadata only on the Alibaba Cloud Model Studio mappings.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro Plus
Run ID: 7a3e91e8-2022-45ce-8574-13376ce7caf5
📒 Files selected for processing (2)
packages/models/src/models/alibaba.tspackages/models/src/models/google.ts
vicovaro
force-pushed
the
feat/deprecation-notices
branch
from
July 28, 2026 16:31
ed55423 to
d515918
Compare
vicovaro
marked this pull request as draft
July 28, 2026 16:39
vicovaro
force-pushed
the
feat/deprecation-notices
branch
from
July 28, 2026 17:07
4279d2e to
e9775b3
Compare
vicovaro
marked this pull request as ready for review
July 28, 2026 17:21
vicovaro
force-pushed
the
feat/deprecation-notices
branch
from
July 28, 2026 17:27
e9775b3 to
f83611a
Compare
Deprecates qwen3-vl-flash, qwen3-coder-plus, qwen3-max, and qwen3.6-max-preview with deactivatedAt set to October 10, 2026 per Alibaba Cloud Service Notice https://www.alibabacloud.com/en/notice/detail?_p_lc=1&id=1950 Co-Authored-By: Claude <noreply@anthropic.com>
Deprecates gemini-2.5-pro, gemini-2.5-flash, and gemini-2.5-flash-lite with deactivatedAt set to October 16, 2026 per Google AI deprecation schedule https://ai.google.dev/gemini-api/docs/deprecations Co-Authored-By: Claude <noreply@anthropic.com>
Align deactivatedAt with Novita's official deprecation date (2026-07-01) for qwen3-vl-8b-instruct, qwen3-next-80b-a3b-thinking, and qwen3-vl-30b-a3b-thinking. The other 11 models in Novita's notice either lack Novita mappings in the catalogue or already had earlier deactivatedAt dates from prior migrations. Reference: https://novita.ai/docs/changelog/25-06-26#jun-25-2026-model-deprecation-notice Co-Authored-By: Claude <noreply@anthropic.com>
vicovaro
force-pushed
the
feat/deprecation-notices
branch
from
July 28, 2026 17:33
f83611a to
1b56e29
Compare
Contributor
Author
|
For the successor of |
Member
E2E results — all green ✅Ran the full gateway e2e suite scoped to every still-active mapping this PR touches: All six Gemini mappings ( All 21 failures were Since Both retirement dates check out: Alibaba's model-depreciation page gives Oct 10 2026 00:00:00 for qwen3.6-max-preview / qwen3-max / qwen3-vl-flash / qwen3-coder-plus, and Enabling auto-merge. |
steebchen
added this pull request to the merge queue
Jul 29, 2026
RATCHAW
pushed a commit
to RATCHAW/llmgateway
that referenced
this pull request
Jul 29, 2026
## Summary Follow-up to the Google deprecation notice for the Gemini 2.5 endpoints on Vertex AI (now "Gemini Enterprise Agent Platform"). **The catalogue already handles the shutdown correctly** — PR theopenco#3288 set `deactivatedAt: 2026-10-16` on both the `google-ai-studio` and `google-vertex` mappings for `gemini-2.5-pro`, `gemini-2.5-flash` and `gemini-2.5-flash-lite`, and `gemini-2.5-flash-image` already carries `2026-10-02`. No `packages/models` changes are needed. What was *not* handled: live docs, guides and a few code-level defaults still recommend Gemini 2.5 as *the* Google model. Those go stale the day the endpoints die. This PR sweeps them to the successors Google names. ## Changes | Area | Before | After | | --- | --- | --- | | `docs/features/{anthropic-endpoint,service-tiers,documents}` | `gemini-2.5-pro` / `gemini-2.5-flash` | `gemini-3.1-pro-preview` / `gemini-3.6-flash` | | `docs/guides/{claude-code,autohand,openclaw}` + `apps/ui` mirrors | same | same | | `token-cost-calculator` popular models | `gemini-2.5-pro`, `gemini-2.5-flash` | `gemini-3.1-pro-preview`, `gemini-3.6-flash` | | models-directory `translation` category | `gemini-2.5-flash`, `gemini-2.5-flash-lite` | `gemini-3.6-flash`, `gemini-3.1-flash-lite` | | `get-provider-endpoint` Vertex fallback model | `gemini-2.5-flash-lite` | `gemini-3.1-flash-lite` | | `/v1/images/generations` OpenAPI example + no-image error hint | `gemini-2.5-flash-image` (retires 2026-10-02) | `gemini-3.1-flash-image-preview` | Also fixes a real bug in the openclaw guide: it advertised the model id `gemini-3-1-pro-preview`, which does not exist in the catalogue (the id is `gemini-3.1-pro-preview`), so anyone copy-pasting that config got a model-not-found. Blog and changelog entries are point-in-time records and were deliberately left untouched. ## Note on the date in the notice The customer email says **October 20, 2026** for the Vertex endpoints. Google's own published lifecycle docs say **October 16, 2026** for these three models on *both* Vertex and the Gemini API, confirmed by the Vertex release note of April 2, 2026 ("The retirement dates for Gemini 2.5 Pro, Gemini 2.5 Flash-Lite, and Gemini 2.5 Flash have been updated to October 16, 2026"). No Google page carries an October 20 date. Our `deactivatedAt` stays at 2026-10-16 deliberately — if the true date is later, we lose four days of availability; if we moved to the 20th and the docs are right, we'd route to a dead endpoint for four days and serve user-facing errors. Conservative wins. ## Test plan - `pnpm format` ✅ - `pnpm build` ✅ (17/17 tasks) - `pnpm test:unit` ✅ (3389 passed, 2 skipped) 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Documentation** * Updated Gemini model examples and recommendations across integration, document-reading, service-tier, and model-selection guides. * Corrected Gemini model identifiers in OpenClaw configuration and available-model references. * **Improvements** * Updated popular and categorized model lists to feature newer Gemini models. * Updated default Google Vertex-compatible model selection. * Refreshed image-related model guidance and error messaging. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Co-authored-by: Claude <noreply@anthropic.com>
This was referenced Aug 9, 2026
steebchen
added a commit
that referenced
this pull request
Sep 16, 2026
## Problem Alibaba Cloud Model Studio retires a batch of legacy models on **2026-10-10**. After that date every public token-pool call to a retired model id fails. Four Alibaba mappings were already marked in #3288; the rest of the batch is still listed as live in our catalogue, and the MCP `generate-image` tool defaults to one of the retiring models. ## Approach Cross-checked the catalogue against the four official Model Studio notices (mainline, snapshot, historical mainline, long-tail, plus the consolidated postponement notice that lists every batch) rather than a summary list, and set `deactivatedAt: new Date("2026-10-10")` on the eight Alibaba mappings that are still active and appear on those lists: | Model | Alibaba's recommended replacement | | --- | --- | | `qwen-omni-turbo` | `qwen3.5-omni-plus` | | `qwen-coder-plus` | Qwen3.6/Qwen3.7 series | | `qwen-image` | `qwen-image-2.0` | | `qwen-image-plus` | `qwen-image-2.0` | | `qwen-image-max` | `qwen-image-2.0` | | `qwen-image-max-2025-12-30` | `qwen-image-2.0` | | `qwen-image-edit-plus` | `qwen-image-2.0` | | `qwen-image-edit-max` | `qwen-image-2.0` | `deactivatedAt` is evaluated against the current time everywhere it is read, so these mappings keep serving until the cutover and then drop out of routing, `/models`, the playground and the cost tools on their own. The remaining retired ids in the notices are either already deactivated in the catalogue (the `2026-07-08` / `2026-07-13` batch, which covers the retired DeepSeek, Kimi and GLM mappings on Model Studio) or not mapped here at all. Because the whole Qwen image-generation line retires, the follow-on edits point at `qwen-image-3.0`: - MCP `generate-image` default model and its example payloads (`apps/gateway/src/mcp/mcp.ts`), plus the matching docs and dashboard copy. - The Alibaba section of the image-generation docs, which is one of the documented exceptions where listing specific models is preferred. ## Notes for the reviewer - The internal announcement's model list is looser than the official notices. Mainline `qwen-max`, `qwen-plus`, `qwen-flash`, `qwen3-vl-plus` and `qwen3.6-plus` do **not** appear on any of the published retirement tables — only their dated snapshots and `-us` variants do, and we do not map those — so they are deliberately left active. Worth a second look if the notice is amended. - After the cutover the catalogue has no active `imageInputRequired` **image** mapping on any provider (the xAI ones are video). Three specs pin models from this batch and will need repointing on or before 2026-10-10: `packages/actions/src/validate-provider-key.spec.ts` (`qwen-image-edit-plus`), `apps/api/src/routes/airside.spec.ts` (`qwen-image-plus`), `apps/gateway/src/chat/managed-credentials.spec.ts` (`qwen-omni-turbo`). They pass today; repointing them now would need a live equivalent that does not exist for the image-edit case, so it is left as a follow-up rather than guessed at. ## Verification `pnpm format`, `pnpm build`, and the `packages/models`, `packages/shared` image-config and `packages/actions` validate-provider-key specs all pass. 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Updated image-generation defaults to `qwen-image-3.0`, with `qwen-image-3.0-pro` shown as an available option. - **Bug Fixes** - Updated MCP tool examples, parameter descriptions, and generated usage guidance to reflect current image-generation model names. - **Documentation** - Refreshed image-generation and MCP reference materials with the latest Alibaba model examples and availability information. - **Deprecations** - The following Alibaba models will no longer accept new requests after October 10, 2026: `qwen-omni-turbo`, legacy Qwen image and image-edit models, and `qwen-coder-plus`. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds
deactivatedAtnotices for models scheduled for retirement.Qwen Models (Alibaba Cloud)
Scheduled for retirement October 10, 2026.
Reference: Alibaba Cloud Service Notice
Gemini Models (Google AI)
Scheduled for retirement October 16, 2026.
Reference: Gemini API Deprecation Schedule
Novita AI Models
Scheduled for retirement July 1, 2026. Three Novita-hosted Qwen models had their
deactivatedAtcorrected to match the official date.11 additional models in Novita's deprecation notice either lack Novita mappings in the catalogue or already had earlier
deactivatedAtdates from prior migrations.Reference: Novita AI Changelog
🤖 Generated with Claude Code
Summary by CodeRabbit