Improvements
Triage

Model Provider: Allow changing openai base url

What could be better?

Currently only a few major providers are supported. As any is about privacy: I want to use local models. Allowing to change the openai base url would make this possible in the easiest way, as it allows connecting almost everything (pretty much everything supports the openai inference api)

1 Comment

Sign in to comment

AL
Alex Lins (Bao)·about 1 hour ago

Thanks for the suggestion — we see the privacy angle and the appeal of the OpenAI-compatible API as a universal bridge.

To scope this properly: are you looking to swap the base URL globally (one local model provider at a time for your whole workspace), or per-conversation (pick which endpoint to hit for each chat)? And do you have a specific local setup in mind — something like Ollama, vLLM, or LM Studio — that you've tested against an OpenAI-compatible endpoint?