Openrouter does not work with Cody

[githubdebugger]

Version

v1.41.1731027960

Describe the bug

Openrouter does not work with Cody

Added this config in settings.json:

{
"provider": "groq", // keep groq as provider
"model": "qwen/qwen-2.5-coder-32b-instruct",
"inputTokens": 128000,
"outputTokens": 8192,
"apiKey": "<api_key>",
"apiEndpoint": "https://openrouter.ai/api/v1/chat/completions"
},

Using groq as the provider with openrouter endpoint (as groq is the provider which needs to be used for OpenAI compatible API).

And this is the response:

Request Failed: HTTP 400 Bad Request: {"error":{"message":"qwen-2.5-coder-32b-instruct is not a valid model ID","code":400}}

Seems like Cody is sending the model as qwen-2.5-coder-32b-instruct instead of qwen/qwen-2.5-coder-32b-instruct ?

Expected behavior

Seems like Cody is sending the model as qwen-2.5-coder-32b-instruct instead of qwen/qwen-2.5-coder-32b-instruct ? and the expected behaviour is to get a response from the endpoint.

Additional context

No response

[linear]

BUGS-711 Openrouter does not work with Cody

[PriNova]

Hey @githubdebugger

Since providing own models via Ollama or third-party services is an experimental feature it might work unexpected.
This ticket is forwarded to the eng team and they will have a look at it.

Sorry for the inconvenience.

[githubdebugger]

This will be available in the next Cody pre-release?

[githubdebugger]

Any updates on this? Will this be available in the official?

[PriNova]

@githubdebugger So far I can't share an update. I am also not aware of any ETA.

Sorry

[tech-and-ai]

Please raise the severity, this is an issue that needs attention!