Jan - Run your own AI

Getting Started - Docs - Changelog - Bug reports - Discord

⚠️ Jan is currently in Development: Expect breaking changes and bugs!

Use offline LLMs with your own data. Run open source models like Llama2 or Falcon on your internal computers/servers.

Jan runs on any hardware. From PCs to multi-GPU clusters, Jan supports universal architectures:

• Nvidia GPUs (fast)
• Apple M-series (fast)
• Apple Intel
• Linux Debian
• Windows x64

Download Jan at https://jan.ai/

Demo

Screenshot: Jan v0.1.3 on Mac M1 Pro, 16GB Sonoma

Quicklinks

• Developer docs (WIP)
• Mobile App shell: App Store | Android
• Nitro Github: Jan's AI engine

Plugins

Jan supports core & 3rd party extensions:

• LLM chat: Self-hosted Llama2 and LLMs
• Model Manager: 1-click to install, swap, and delete models with HuggingFace integration
• Storage: Optionally save conversation history and other data in SQLite
• 3rd-party AIs: Connect to ChatGPT, Claude via API Key (in progress)
• Cross device support: Mobile & Web support for custom shared servers (in progress)
• File retrieval: User can chat with docs
• Multi-user support: Share a single server across a team/friends (planned)
• Compliance: Auditing and flagging features (planned)

Nitro (Jan's AI engine)

In the background, Jan runs Nitro, an open source, C++ inference engine. It runs various model formats (GGUF/TensorRT) on various hardware (Mac M1/M2/Intel, Windows, Linux, and datacenter-grade Nvidia GPUs) with optional GPU acceleration.

See the open source Nitro codebase at https://nitro.jan.ai.

Troubleshooting

As Jan is development mode, you might get stuck on a broken build.

To reset your installation:

1. Delete Jan Application from /Applications

2. Clear cache: rm -rf /Users/$(whoami)/Library/Application\ Support/jan-electron OR rm -rf /Users/$(whoami)/Library/Application\ Support/jan

---

Contributing

Contributions are welcome! Please read the CONTRIBUTING.md file

Pre-requisites

• node >= 20.0.0
• yarn >= 1.22.0

Instructions

Note: This instruction is tested on MacOS only.

1. Clone the Repository:

   git clone https://github.com/janhq/jan
   git checkout DESIRED_BRANCH
   cd jan
   

2. Install dependencies:

   yarn install
   
   # Packing base plugins
   yarn build:plugins
   

3. Run development and Using Jan Desktop

   yarn dev
   

   This will start the development server and open the desktop app. In this step, there are a few notification about installing base plugin, just click OK and Next to continue.

For production build

# Do step 1 and 2 in previous section
git clone https://github.com/janhq/jan
cd jan
yarn install
yarn build:plugins

# Build the app
yarn build

This will build the app MacOS m1/m2 for production (with code signing already done) and put the result in dist folder.

License

Jan is free, open core, and Sustainable Use Licensed.

Acknowledgements

Jan builds on top of other open-source projects:

• llama.cpp
• TensorRT
• Keycloak Community (Apache-2.0)

Contact

• Bugs & requests: file a Github ticket
• For discussion: join our Discord here
• For business inquiries: email hello@jan.ai
• For jobs: please email hr@jan.ai