Enhancement:

[cauvang32]

What features would you like to see added?

Add GPT-4o mini, o1-preview and o1-mini support (or even more) for Github Model.

More details

Github Models has many different types of models so I hope your team can add that functionality, I have been waiting for that for a long timeee.

Which components are impacted by your request?

No response

Pictures

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[danny-avila]

Have you tried adding it as a custom endpoint?

https://www.librechat.ai/docs/quick_start/custom_endpoints

[cauvang32]

Yes I've tried that, however I get this error when using o1-mini and o1-preview
2024-12-07 13:53:33 info: Server listening on all interfaces at port 3080. Use http://localhost:3080 to access it
{"level":"info","message":"[Login] [Login successful] [Username: [email protected]] [Request-IP: 172.19.0.1]","timestamp":"2024-12-07T13:53:49.000Z"}
2024-12-07 13:53:50 error: Failed to fetch models from Github OAI API
Something happened in setting up the request. Error message:
Cannot read properties of undefined (reading 'map')
2024-12-07 13:54:03 warn: [OpenAIClient.chatCompletion][stream] API error
2024-12-07 13:54:03 error: [handleAbortError] AI response error; aborting request: 400 Unsupported value: 'stream' does not support true with this model. Only the default (false) value is supported.
2024-12-07 13:54:16 warn: [OpenAIClient.chatCompletion][stream] API error
2024-12-07 13:54:16 error: [handleAbortError] AI response error; aborting request: 400 Unsupported value: 'stream' does not support true with this model. Only the default (false) value is supported.

[harrisonhxy]

Have you implemented the GitHub models o1-mini and o1-preview in librechat? 4o as well as 4o-mini etc. of GitHub models work fine, how should the o1 series models be called?
By the way, I'm sure I've got access to GitHub models o1-mini and o1-preview!

[cauvang32]

I'm using custom api endpoint in Librechat.yaml like this:

version: 1.1.4
cache: true
endpoints:
  custom:
    - name: "Github OAI"
      apiKey: "my_github_personal_token"
      baseURL: "https://models.inference.ai.azure.com"
      models:
        default: ["gpt-4o","gpt-4o-mini","o1-mini","o1-preview"]
        fetch: true
      titleConvo: true
      titleModel: "current_model"
      summarize: false
      summaryModel: "current_model"
      forcePrompt: false
      modelDisplayLabel: "OpenAI-Github"

[cauvang32]

However I don't know how to remove system prompt or steam respond setting in o1-mini and o1-preview calll:(

[cauvang32]

yeah and I knew that o1 model now didn't support system prompt and steam respond so how I can remove that from the request?

[cauvang32]

@harrisonhxy do you have any idea?

[harrisonhxy]

You mean not using the o1 series of GitHub models right? May I ask if your api key source is official, or is it a third-party proxy like transit api key (e.g., one api/new api)
Because I'm using a third-party proxy's api key that is aggregated using the new api, I just need to exclude the GitHub models channel's o1 series in the new api

[cauvang32]

No, I'm using the official github models however the default github model section don't support o1-mini and o1-preview so I use the custom endpoint: https://models.inference.ai.azure.com (official github models api endpoint)

[harrisonhxy]

So why can't it be removed from the request?Just don't put the o1 series model in the custom endpoint, right?