Jan is a free, open-source AI assistant that transforms any desktop (Mac, Windows, Linux) into a private AI powerhouse. It runs locally using popular models like Llama3, Gemma, or Mistral—no internet needed—keeping your conversations and data completely private. Plus, if you want it, Jan can connect to cloud models like OpenAI or Anthropic too.
Key Features
Offline-first Local Chat: Loads AI models on your own device—chat without internet, preserving privacy.
Model Hub & Cloud Options: Choose local models (Llama3, Mistral, Gemma) or plug in API keys to use cloud models.
OpenAI-compatible API Server: Run a local API at localhost:1337—ideal for developers and tool integrations.
Highly Extensible: Add extensions, tools, and future memory‑based assistants via a plugin‑style system.
Use Cases
Personal Chats & Brainstorming: Ask questions, plan ideas, and write—all offline.
Code Assistance & Tool Integration: Use Jan’s API to beef up editors or automate tasks locally.
Data & Document Analysis: Load documents to chat with your files and surface insights—future version.
Technical Specifications
Supported OS & Hardware: macOS ≥ 13, Windows 10+, Linux (glibc ≥ 2.27); supports NVIDIA (CUDA), AMD/Intel GPUs (Vulkan), Apple M-series.
Powered by Cortex Engine: Under the hood, Cortex runs local models with OpenAI‑style API compatibility.
AGPLv3 Open‑Source License: Fully open-source, permissive for personal and community use.