Repository navigation
API Key 401 Invalid API key — muse-spark-1.2-contributor models list works but chat rejected #46169
Description
Description
I am located in Taiwan, which is not among Meta's restricted territories for API access.
API Key (sk-Nkp...) successfully calls the /models endpoint (HTTP 200) and correctly returns the model list including muse-spark-1.2-contributor.
However, all chat completion / responses endpoints return 401 Invalid API key regardless of which model is used:
muse-spark-1.2-contributor→ 401Invalid API keymuse-spark-1.2-contributor-free→ 401Model muse-spark-1.2-contributor-free is not supported(confirmed this is not a valid model name)gpt-5.3-codex→ 401Invalid API key
Endpoints tested:
https://opencode.ai/zen/go/v1/chat/completions→ 401https://opencode.ai/zen/go/v1/responses→ 401
Test time: 2026-08-31 ~01:00 +08:00 (UTC+8)
Relevant settings:
- "Allow models that use request data for training" is enabled (this is a data privacy setting and likely unrelated to API key validation)
Expected behavior:
API Key with /models access should be able to call chat completion endpoints.
Actual behavior:
All chat-related endpoints return 401 Invalid API key.
Request:
Please confirm whether this API key has chat completion / responses permissions enabled, or if a new key needs to be generated.
Plugins
No response
OpenCode version
No response
Steps to reproduce
No response
Screenshot and/or share link
Operating System
No response
Terminal
No response
Activity
github-actions commentedon Aug 29, 2026
This issue might be a duplicate of existing issues. Please check:
/modelsbut rejected on chat endpointsIf your issue is different from the above, please add more detail to distinguish it (e.g., account type, subscription plan, when the key was generated).
felixzsh commentedon Sep 10, 2026
im in mexico (not restricted) and i get this error sporadically: Error: Error from provider (Console Go): Upstream request failed: [invalid_api_key] Unauthorized
Tarquinen commentedon Oct 20, 2026
Thanks for the report! Closing as a duplicate of #50885, which has the answer: #50885 (comment)
In short: in OpenCode 2.0, run
/connect, chooseopencode, and pick the workspace that has Go when signing in. Then select a model marked as Go. To use an API key instead, create one under a service account in the console and add it with/connect→opencode go.If this still fails on 2.0, please open a new issue with your version.
— from 𝕺𝖕𝖊𝖓𝕮𝖔𝖉𝖊