[Feature] provide a unified API to call a model chat completion

[ZuidVolt]

Issue Type

• Bug report
• Feature request
• Documentation update
• Other (please specify)

Description

once we support more model API's than just Gemini or want to support more than one session of Gemini running (concurrent execution) we will need a model API router data structures and a model queue data structures

info

celeryq might be a good choice for the model queue data structures