Make it compatible with deepinfra

I am trying to use thsi plugin with deepinfra which offers llama3b-70b on the cheap and a openAI compatible api, how ever its api endpoints has "openai" in the middle like this

https://api.deepinfra.com/v1/openai/chat/completions

And current code hard-coded to add `/v1/chat/completions` path to api base address, and this makes it not able to make api calls correctly to deepinfra api endpoint.

Would it be possible to make "v1" part of api base url? I know this is a breaking change, but it does not make a lot of sense to hard-code an api version in request paths, right?

Start by locating the request code that hard-codes `/v1/chat/completions` onto the API base address. Check how the base URL is configured and verify that allowing the versioned path in the base address works with the DeepInfra endpoint without breaking existing API calls.

Written by the indexing model from the issue text.