Conversation
…nses API Upstream serves Muse Spark only on /v1/responses; without a registry targetFormat entry the executor falls back to /chat/completions and the upstream returns 500 (same class as diegosouzapw#12196). Add base + minimal/low/medium/ high/xhigh effort aliases (verified via `opencode models opencode-go --refresh --verbose`) with targetFormat openai-responses, register the effort tiers for alias parsing, and cover with regression tests. Fixes diegosouzapw#12674
4 of 5 tasks
maxmad64bis
pushed a commit
to maxmad64bis/OmniRoute
that referenced
this pull request
Sep 12, 2026
… Zen, OpenCode and OpenCode Go Muse Spark 1.3 is only served on the Responses API, like 1.2. None of the 1.3 ids were registered with targetFormat "openai-responses", so the executor fell back to /chat/completions and the upstream answered 500. - opencode-go: muse-spark-1.3-contributor and its minimal/low/medium/high/xhigh effort aliases, plus the 1.3 effort tiers in the executor (from diegosouzapw#12675). - opencode-zen and opencode: muse-spark-1.3 and muse-spark-1.3-contributor-free. - All 1.3 entries declare contextLength 1048576 and maxOutputTokens 131072, matching the published OpenCode catalog and the 1.2 entries, instead of the 200000 provider default. Builds on diegosouzapw#12675, and on the same fix proposed in diegosouzapw#12973 and diegosouzapw#13111. Co-authored-by: bacnh85 <mail@bacnh.com>
Owner
|
Thanks — this is a very thorough writeup (the upstream curl reproduction really nails the root cause), and you get credit as the original reporter of #12674. Your changelog fragment is also the only one of the three PRs on this issue that included one — we're planning to carry it forward. The maintainer's follow-up on the issue asked to also cover the Triage note: this is the review recommendation — the close itself happens only after the maintainer's per-PR sign-off (and, where a superseding PR is named, after it has landed). Nothing is being closed by this comment. |
diegosouzapw
pushed a commit
that referenced
this pull request
Sep 15, 2026
… Zen, OpenCode and OpenCode Go (#13471) Registers Muse Spark 1.3 (with its effort aliases) on OpenCode Zen, OpenCode and OpenCode Go with `targetFormat: openai-responses` and a 1M context window, so the model no longer falls back to `/chat/completions` (#12674, #12698). Superset of #12675, #12973 and #13111, whose authors are credited in the PR. Validated first on the combined board of all 38 PRs of this batch (10 merged as-is, 28 after the maintainer rework) on top of release/v3.8.51 c0f92ec: typecheck:core, check:open-sse-typecheck and check:dashboard-typecheck clean; ESLint clean on every changed file; file-size (rebaselined for the combined growth), complexity, cognitive-complexity, changelog-integrity, docs-counts, docs-sync, migration-numbering and i18n new-key gates green; 735 focused node:test cases with the only batch-caused failure (a flag-count assertion) fixed. Then re-validated alone on the fresh release tip right before this merge: ESLint on the changed files, typecheck:core, check:open-sse-typecheck, the file-size/complexity/changelog gates and this PR's own tests. Thanks @maxmad64bis!
Owner
|
Thanks for catching this! Muse Spark 1.3 routing through the Responses API landed via #13471, which covers OpenCode Zen, OpenCode and OpenCode Go plus the effort aliases and credits the parallel reports. The registry entries, |
muhamadgalihsaputra
pushed a commit
to niyatna/NiyatnaRoute
that referenced
this pull request
Sep 27, 2026
… Zen, OpenCode and OpenCode Go (diegosouzapw#13471) Registers Muse Spark 1.3 (with its effort aliases) on OpenCode Zen, OpenCode and OpenCode Go with `targetFormat: openai-responses` and a 1M context window, so the model no longer falls back to `/chat/completions` (diegosouzapw#12674, diegosouzapw#12698). Superset of diegosouzapw#12675, diegosouzapw#12973 and diegosouzapw#13111, whose authors are credited in the PR. Validated first on the combined board of all 38 PRs of this batch (10 merged as-is, 28 after the maintainer rework) on top of release/v3.8.51 d61b804: typecheck:core, check:open-sse-typecheck and check:dashboard-typecheck clean; ESLint clean on every changed file; file-size (rebaselined for the combined growth), complexity, cognitive-complexity, changelog-integrity, docs-counts, docs-sync, migration-numbering and i18n new-key gates green; 735 focused node:test cases with the only batch-caused failure (a flag-count assertion) fixed. Then re-validated alone on the fresh release tip right before this merge: ESLint on the changed files, typecheck:core, check:open-sse-typecheck, the file-size/complexity/changelog gates and this PR's own tests. Thanks @maxmad64bis!
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.
Fixes #12674.
Upstream (
opencode.ai/zen/go) serves Muse Spark only on the Responses API (/v1/responses); direct/v1/chat/completionsreturns 500 for both 1.2 and 1.3 (verified same key, same server). The 1.2 family has registrytargetFormat: "openai-responses"entries, but 1.3 (upstream release 2026-09-02) was never added, soresolveOpencodeTargetFormat()falls back to"openai"and the request 500s (retried 3x, account cooldown,(reset after 3s)).Changes
open-sse/config/providers/registry/opencode/go/index.ts: 1.3 base + minimal/low/medium/high/xhigh effort aliases (no max), eachcontextLength 1048576 / maxOutputTokens 131072 / supportsReasoning+vision+audio+video / targetFormat openai-responses, mirroring the 1.2 block. Specs verified viaopencode models opencode-go --refresh --verbose(same effort set, limits, and $0.10/$0.20 pricing as 1.2).open-sse/executors/opencode.ts:EFFORT_TIERS["muse-spark-1.3-contributor"] = ["minimal","low","medium","high","xhigh"]so suffixed aliases parse and get reasoning_effort injection.tests/unit/opencode-go-muse-spark-1-3-12674.test.ts: registry entries, executor URL resolution, parseEffortLevel (incl. max→null), Go-only scoping.Verification
opencode-go-effort-aliases-8353,opencode-muse-spark-responses-10867,opencode-target-format-alias-11045,opencode-muse-spark-min-output,opencode-session-fingerprint-headers-10571: 64/64 pass.check-provider-consistency: OK.Out of scope (noted in #12674): bare
muse-spark-1.3-contributorcombo path also triescommand-code/meta/...→ 502 empty response; may need the same treatment in that provider.