idea: Support multiple LLMs & embedding models through server

[Capsar]

Problem Statement

As a developer, I am also interested in hosting embedding models locally, besides the usual coder LLMs used in the IDE.

Furthermore, there are other types of models, such as image-to-text for OCR / or document extraction which are also lacking support at the moment.

Feature Idea

• Be able to host multiple LLMs & Embedding models through the server.
  • https://huggingface.co/jinaai/jina-embeddings-v3
• Be able to host image-text-to-text models through the server.
  • https://huggingface.co/ds4sd/SmolDocling-256M-preview, https://huggingface.co/starvector/starvector-8b-im2svg

[DKostk]

Greetings!
Is it possible to add a function when creating a prompt, for example, using @ or # with the name of the local AI version?
Thus, different ais will perform all tasks according to prompta.