[Enhancement]: Restrict Some OpenAI Models via Environment Variables

[LemonDouble]

What features would you like to see added?

According to the OpenAI documentation0 (Link), I can set the OPENAI_MODELS environment variable to configure which models are available.

However, I have the following use cases:

1. I want to retain the automatic model list fetching via the API (I want to try new models as soon as they come out).
2. Among those, I’d like to exclude models that are “excessively expensive” (such as o1-pro) or models that aren’t suited for chat purposes (like gpt-4o-transcribe).

More details

I believe this can be achieved by modifying the section found in this Line (Link).

If you think this is a useful feature, I'd be happy to open a PR.

Thank you for creating such an awesome project!

Which components are impacted by your request?

Endpoints

Pictures

Some API (Like gpt-4o-mini-transcription) are literally not work.

Code of Conduct

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