Enable to specify OpenAI API URL to enable compatible services

[frankhommers]

I would like to use the Microsoft Azure OpenAI API instead of the OpenAI API.
I believe only the URL needs to change, everything else should be the same.

It would be sufficient to specify an API URL when selecting AI Provider: OpenAI (ChatGPT)

<3 the project

[clusterzx]

This is a thing I will also implement soon 🙂

Thank you for kind words.

[nickwilkie]

One step further, a lot of us homelabbers use a local OpenAI API to run open source models rather than Ollama. Personally I use VLLM since it's easier to keep a single model loaded into VRAM rather than loading/unloading like Ollama tends to do. It would be great to be able to specify any address for OpenAI as well as the model name rather than the dropdown with 3 options that's currently implemented.

It would also be easier to upkeep I think, since Ollama itself can serve an OpenAI API, so you could do a single configuration option.

[tajchert]

Yes, this would be extremely helpful to also use proxy services such as OpenRouter.ai

It offers OpenAI compatible API (under https://openrouter.ai/api/v1) and allows to use hundreds of models with a single payment and API key.
However it would also require to be able to specify not only a API URL as open input but as well model id as in openrouter there is provider as a prefix for model ex. openai/chatgpt-4o-latest

[clusterzx]

Done my friends :)