idea: cortex should auto-reload model when start chat completion

[vansangpfiev]

Current Process Overview

At present, initiating a chat completion requires sending a start model request. However, when switching engines, Jan is unable to determine the current state, which prevents the model from starting effectively.

Proposed Improvement

To streamline this process, Cortex should implement functionality to automatically save the previous configuration and initiate the model without manual intervention. This enhancement will ensure a smoother transition between engines and improve overall efficiency.