Skip to content

The API endpoint for muse-spark-1.2-contributor is unreachable #44627

Closed as not planned

Description

@superuser7777

Description

I received a response without any issues using the opencode CLI. (This is the standard version, not the free version.)
Muse Spark 1.2 Contributor · 3.1s

Does anyone know the cause and how to fix it? Is this a server-side issue?

  • location: japan (This country is not subject to API service restrictions.)
  • os: windows 11 enterprise latest
  • terminal: windows terminal, wezterm (latest)
  • shell: pwsh(latest)
  • curl 8.21.0 (Windows) libcurl/8.21.0 Schannel zlib/1.3.2 WinIDN WinLDAP

ocgo official endpoint: https://opencode.ai/zen/go/v1/chat/response

 $KEY = (Get-Content "D:\OpenCode_api_key.txt" -Raw).Trim(); '{"model":"muse-spark-1.2-contributor","messages":[{"role":"user","content":"say:a"}]}' | Set-Content -Encoding ascii "$env:TEMP\ocbody.json"; curl.exe -sS -i --max-time 30 -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" -d "@$env:TEMP\ocbody.json" https://opencode.ai/zen/go/v1/chat/response
HTTP/1.1 404 Not Found
Date: Mon, 24 Aug 2026 05:57:50 GMT
Content-Type: text/html
Transfer-Encoding: chunked
Connection: keep-alive
Cf-Placement: remote-ORD
Server: cloudflare
CF-RAY: a3001f2c5c1d5ecd-PDX

<ocgo 404 not found page html raw>

endpoint https://opencode.ai/zen/go/v1/chat/completions

 $KEY = (Get-Content "D:\OpenCode_api_key.txt" -Raw).Trim(); '{"model":"muse-spark-1.2-contributor","messages":[{"role":"user","content":"say:a"}]}' | Set-Content -Encoding ascii "$env:TEMP\ocbody.json"; curl.exe -sS -i --max-time 30 -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" -d "@$env:TEMP\ocbody.json" https://opencode.ai/zen/go/v1/chat/completions
HTTP/1.1 500 Internal Server Error
Date: Mon, 24 Aug 2026 05:55:26 GMT
Content-Type: text/plain;charset=UTF-8
Content-Length: 75
Connection: keep-alive
Cf-Placement: remote-ORD
Server: cloudflare
CF-RAY: a3001bad3a95e653-DEN

{"type":"error","error":{"type":"error","message":"Internal server error"}}

mimo-v2.5 is 200 ok

$KEY = (Get-Content "D:\OpenCode_api_key.txt" -Raw).Trim(); '{"model":"mimo-v2.5","messages":[{"role":"user","content":"only say->a"}],"max_tokens":5}' | Set-Content -Encoding ascii "$env:TEMP\ocbody.json"; curl.exe -sS -i --max-time 30 -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" -d "@$env:TEMP\ocbody.json" https://opencode.ai/zen/go/v1/chat/completions
HTTP/1.1 200 OK
Date: Mon, 24 Aug 2026 07:32:24 GMT
Content-Type: application/json
Content-Length: 479
Connection: keep-alive
Cf-Placement: remote-ORD
Server: cloudflare
CF-RAY: a300a98a48027537-SEA

Plugins

No response

OpenCode version

latset

Steps to reproduce

No response

Screenshot and/or share link

No response

Operating System

windows11 latest

Terminal

windows terminal, wezterm latest

Activity

  1. github-actions commented on Aug 24, 2026

    @github-actions
    Contributor

    This issue might be a duplicate of existing issues. Please check:

  2. DevonGithub commented on Aug 24, 2026

    @DevonGithub

    I reproduced this and isolated the failing path further.

    Environment:

    • Windows 11
    • OpenCode Go
    • Model: muse-spark-1.2-contributor
    • OpenCodex 2.32.0
    • Codex Desktop
    • Same OpenCode Go API key for all tests

    Results:

    1. Muse works normally inside the OpenCode app.

    2. Direct request to the Responses API works:

    POST https://opencode.ai/zen/go/v1/responses

    with:

    model: muse-spark-1.2-contributor

    Result:

    HTTP 200
    status: completed

    So the model, account, API key and OpenCode Go access are all working.

    1. The same model routed through the Chat Completions protocol fails consistently:

    /v1/chat/completions

    Result:

    HTTP 500
    Internal server error

    1. OpenCodex debug confirms the failing request is using:

    adapter: openai-chat
    model: muse-spark-1.2-contributor
    stream: true

    1. After a completely fresh OpenCodex 2.32.0 reinstall, other models through the same setup work correctly:
    • GPT-5.6 Luna: 200
    • DeepSeek V4 Flash: 200
    • Gemini 3.7 Flash: 200
    • Muse Spark 1.2 Contributor: 500

    This therefore appears isolated to Muse + the Chat Completions API path.

    The OpenCode Go documentation currently lists Muse Spark 1.2 Contributor under /v1/responses, and our direct test confirms that endpoint works.

    Possible fixes:

    • Make muse-spark-1.2-contributor work correctly through /v1/chat/completions, or
    • Return a clear 4xx unsupported-protocol error instead of HTTP 500 when Muse is called through Chat Completions, and
    • Ensure clients can reliably discover that this model requires the Responses API.

    This also seems related to the existing Muse Chat Completions streaming/schema issues, but in this case the request fails immediately with HTTP 500 before a usable response is returned.

  3. Tao-Yida commented on Aug 27, 2026

    @Tao-Yida

    I reproduced this and isolated the failing path further.

    Environment:

    • Windows 11
    • OpenCode Go
    • Model: muse-spark-1.2-contributor
    • OpenCodex 2.32.0
    • Codex Desktop
    • Same OpenCode Go API key for all tests

    Results:

    1. Muse works normally inside the OpenCode app.
    2. Direct request to the Responses API works:

    POST https://opencode.ai/zen/go/v1/responses

    with:

    model: muse-spark-1.2-contributor

    Result:

    HTTP 200 status: completed

    So the model, account, API key and OpenCode Go access are all working.

    1. The same model routed through the Chat Completions protocol fails consistently:

    /v1/chat/completions

    Result:

    HTTP 500 Internal server error

    1. OpenCodex debug confirms the failing request is using:

    adapter: openai-chat model: muse-spark-1.2-contributor stream: true

    1. After a completely fresh OpenCodex 2.32.0 reinstall, other models through the same setup work correctly:
    • GPT-5.6 Luna: 200
    • DeepSeek V4 Flash: 200
    • Gemini 3.7 Flash: 200
    • Muse Spark 1.2 Contributor: 500

    This therefore appears isolated to Muse + the Chat Completions API path.

    The OpenCode Go documentation currently lists Muse Spark 1.2 Contributor under /v1/responses, and our direct test confirms that endpoint works.

    Possible fixes:

    • Make muse-spark-1.2-contributor work correctly through /v1/chat/completions, or
    • Return a clear 4xx unsupported-protocol error instead of HTTP 500 when Muse is called through Chat Completions, and
    • Ensure clients can reliably discover that this model requires the Responses API.

    This also seems related to the existing Muse Chat Completions streaming/schema issues, but in this case the request fails immediately with HTTP 500 before a usable response is returned.

    I also reproduced this 500 Error on Codex via OpenCodex v2.33.0. For me, muse-spark-1.2-contributor on OpenCode Go returns 500 in log yet showing "we're currently experiencing high demand, which may cause temporary errors", and muse-spark-1.2-free on OpenCode Zen returns 401 Invalid API Key. However, on OpenCode TUI both two models work well. During the first day of Muse I found it glitches on Codex because of the different schema it uses. Hope they can fix it soon because my quota is running out.

  4. superuser7777 commented on Aug 28, 2026

    @superuser7777
    Author

    Is the OpenCode team looking at this issue?
    Several days have passed, but the situation hasn't changed at all.
    To help isolate the problem, I'd like to know whether it's an OpenCode issue or a Meta issue.

  5. Tarquinen commented on Oct 20, 2026

    @Tarquinen
    Member

    Thanks for the report! Muse Spark is served on the /responses endpoint, not /chat/completions or /chat/response. Each Go model is available on the endpoint listed for it in the Go docs: https://opencode.ai/v2/docs/console/go#endpoints

    As confirmed in the comments, requests to /responses work. Note that Muse is also only available in some regions. Closing this issue.

    — from 𝕺𝖕𝖊𝖓𝕮𝖔𝖉𝖊

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions