Create an integration with Jan.ai

Created on 15 October 2024, about 1 month ago

Problem/Motivation

Jan.ai is much like LLM Studio but with a greater privacy focus.
https://jan.ai/
https://github.com/janhq/jan

Can Jan.ai be a provider for the AI module?

Steps to reproduce

Proposed resolution

Remaining tasks

User interface changes

API changes

Data model changes

✨ Feature request
Status

Active

Version

1.0

Component

Providers

Created by

πŸ‡ΊπŸ‡ΈUnited States SocialNicheGuru

Live updates comments and jobs are added and updated live.
Sign in to follow issues

Comments & Activities

  • Issue created by @SocialNicheGuru
  • πŸ‡©πŸ‡ͺGermany marcus_johansson

    It will not be in the AI core module, because we will be moving out all providers from core into contrib modules now :)

    However if you want to either find people to do this or support it or want to support it yourself, I have no problems doing the initial development for you. I can however not support it, I have to many projects to support already. So, if I am to build it I need to know that there are at least two maintainers ready to take any support requests and bug requests coming in.

    One place to look for other people to help you out are on the channels #ai and #ai-provider-maintainers.

    Since it will not be a feature of the AI module, I'll switch this to a support request so we can see if someone else is looking for this provider and ready to help.

  • πŸ‡ΊπŸ‡ΈUnited States SocialNicheGuru

    Point me in the direction. what woudl it look like. should I look at the LLM Studio provider and make modifications based on the Jan.ai api?

  • πŸ‡©πŸ‡ͺGermany marcus_johansson

    Without knowing jan.ai, LMStudio I think is the closest one. As long as its OpenAI compatible, which it looks like, it should be fairly easy and mostly copy and paste (see https://jan.ai/docs/remote-models/generic-openai).

    You would need some way of either getting the models, I couldn't find anything on a quick google. If that is not available, the user has to fill in the models themselves, there is a generic way of doing this, you can see in Huggingface provider (or Azure provider).

Production build 0.71.5 2024