LiteLLM ModelClient Integration

[PhiBrandon]

Description & Motivation

Given the landscape of Inference providers it seems like a good idea to have a way to interact with then using a relatively 'proven' library. Developing a ModelClient for LiteLLM would allow users to interact 100+ LLMs with more coming online everyday.

Pitch

There was a previous attempt at this, that has since been updated. https://github.com/SylphAI-Inc/AdalFlow/tree/litellm. This was implemented prior to optimization updates. I want to add this issue here for communication on this ModelClient.

Provider POC

Needs active outreach.

Additional context

No response

[liyin2015]

the litellm team is not responsive

@PhiBrandon do you want to create a pr? Do users really find litellm useful? Let's wait to see if people will upvote for it.

I feel it will be more secure to use each provider's sdk (and often they are more supportive on the integration) and easier to maintain rather than introducing a 3rd party proxy

Note: Do not create a pr until we put up "ready-for-PR" label

[PhiBrandon]

Based on discord vote this wouldn't be the best use of time at the moment.