Prerequisites

• I have searched existing issues and discussions to avoid duplicates

Problem to solve

Currently for ChatCompletion request we use openai compatible endpoint provided by gemini. This is leading to information loss as Generation Request has lot more parameters than Chat Completion.

Proposed solution

Add necessary converters for using gemini api request/response format. Remove dependency on openai compatible endpoints.

Alternatives considered

No response

Area(s)

Core (Go)

Additional context

No response