[Feature Request] Support OpenAI-compatible providers that do not publish the models endpoint

What's happening?

I have a custom provider that is OpenAI-compatible for chat responses but does not publish the model list, so the current workaround is to populate the models list from the UI by choosing a provider like "LLM API", then editing the API endpoint. Finally, the model ids need to be changed by editing the app-models entry in local storage.

This is rather messy, so a way for users to add/edit manual entries to the model list is desired.

Where does this happen?

Self-deployed from GitHub

Impact on your workflow

Medium - Workaround exists

Environment (if applicable)

No response

Additional context

No response