Create way for devs to utilize private models in build process

[lxy009]

while a private model with an API in front hosted somewhere else can utilize the same integration as the proprietary models, it might be nice to deploy the model alongside the application in one go and on the same server.

[kalexander]

we can already do this with the https://github.com/open-build/BabbleBeaver/pull/19 code using ollama.. what if we change the requirements to allow models that are compatible with the OpenAI API. This would give us the following:

• satisfying this ticket..
• satisfying https://github.com/open-build/BabbleBeaver/issues/20 as hugging face also as a Text Generation Inference toolkit that serves up hugging face models via the openai protocol
• allowing devs to use other services like
  • openrouter.ai
  • together.ai
  • cloudflare - who knew CloudFlare was in the game?!?!
  • and i'm sure many more...