langchain to avoid vendor lock in

[belda]

Hi have you considered using https://github.com/langchain-ai/langchain ?

It would be a nice addition to easily replace the LLM driving it and soon we can run completely opensource model locally.

[TheR1D]

It is possible to run other LLMs locally using LocalAI or other OpenAI compatible local backends. In the config file we can point sgpt requests to localhost by changing OPENAI_BASE_URL.

[belda]

Omg, great to find out about LocalAI!