[FEATURE] Local LLM?

[Mistborn-First-Era]

Did you have plans for local?
Could you set it up to run with a local LLM/software. I would prefer to keep my data private, have offline capabilities, and avoid any costs associated with online resources.
Allowing the model to be selected from one of the many huggingface repos would mean that VRAM compatible models could be selected on a user by user basis.