Support Ollama, LM Studio, Llama.cpp, and other model providers/servers

[wbste]

Currently, the endpoint for OpenAI sends a request to /responses, which is an OpenAI only endpoint AFAIK. Most "OpenAI-compat" servers rely on /chat/completions, and the client app maintains message state.

See: https://github.com/ollama/ollama/blob/main/docs/openai.md or https://lmstudio.ai/docs/app/api/endpoints/openai

It looks like the easiest thing is to add a few providers:

https://ai-sdk.dev/providers/community-providers/ollama, although there is an warning at the top about the embeddings.
https://ai-sdk.dev/providers/openai-compatible-providers/lmstudio

I think simply adding @ai-sdk/openai-compatible would take care of Ollama, LM Studio, Llama.cpp, and many other model providers.

[wbste]

Seems I had the wrong keyword when I searched. Found this: #419
Closing to track over there!