Separate profile for Autocomplete

[isupovs]

I have two computers running local LLMs. On the first machine, I run a large model that occupies all available VRAM.

As I understand it, the Autocomplete feature currently supports only Mistral-based models, which is fine. However, I would like to run Autocomplete on a second machine, while keeping the main assistant on the first one.

At the moment, there is no way to configure Autocomplete to use a separate profile or a different endpoint/host. This makes it impossible to split workloads between machines.

[emilieschario]

@isupovs you should be able to do this- here's an instructional video: https://www.youtube.com/watch?v=iNioZeF4-yY

[isupovs]

It's not what I need. I don't want to use a cloud provider. I want it to run locally.

[oldjs]

Yes, that's fair