Ollama vs. OpenAI: Shouldn't we just have OpenAI support (as Ollama supports OpenAI)?

[ChristianWeyer]

Hey all,

first, thanks for the great lib! :-)

Looking at https://ai.pydantic.dev/api/models/openai/ and https://ai.pydantic.dev/api/models/ollama/ I am wondering why not just have the OpenAI one and provide e.g. a base URL property?

As Ollama implements the OAI API, this seems like a streamlined way to do. Then we can use OpenAI() for both real OpenAI and local / remote inference engines like Ollama.

Thoughts?

[Navanit-git]

i yess thats what I was talking about.

[samuelcolvin]

I think it's worth having a dedicated model, even if it's just setting the base url.

Also, look like we'll need more specific support for Ollama, see #242.

[ChristianWeyer]

Then we should at least have the base URL setting in the OpenAI model - OK @samuelcolvin?