Jan

Key Features of Jan

• 100% Offline AI
  Run large language models completely on your local machine — no internet required.

• Open Source & Transparent
  Fully open-source with active development on GitHub.

• Cortex Engine
  Jan is powered by Cortex, a local-first AI runtime optimized for speed and flexibility.

• Cross-Platform Support
  Compatible with Windows, macOS (Intel & M-series), and Linux (Debian-based).

• Model Library
  Easily run popular LLMs like Llama, Gemma, Mistral, and Qwen.

• Remote API Access
  Optionally connect to services like Groq and OpenRouter if needed.

• OpenAI-Compatible API Server
  Serve local models with an API compatible with OpenAI's — great for developers.

• File Chatting (Experimental)
  Interact with documents and files using local models.

• Highly Customizable
  Modify behavior, models, UI, and more via extensions.

• Future Assistant & Memory Support
  Upcoming features include personalized assistants and long-term memory.

• Thriving Community
  Over 28K GitHub stars, 3.1M+ downloads, and active discussions on Discord.

Meet Jan: A ChatGPT Alternative That Runs 100% Offline

Imagine having a powerful AI chatbot like ChatGPT, but one that runs entirely on your own device, with no internet connection required. That’s what Jan delivers: a privacy-first, open-source AI assistant that works offline, out of the box.

Built for users who care about control, transparency, and performance, Jan makes local AI not just possible, but easy.

Why Jan Exists

Jan was created with a clear vision: AI should be open, local-first, and user-owned. For too long, using advanced AI tools has meant giving up privacy and sending your data to remote servers. Jan changes that. You download it once, run it on your machine, and everything, from your conversations to your models, stays with you.

No cloud. No surveillance. No strings attached.

Powered by Cortex: AI That Runs Anywhere

At the heart of Jan is Cortex, a lightweight, embeddable AI engine optimized to run large language models (LLMs) locally. Whether you’re on a Windows PC, a Mac with M-series silicon, or a custom multi-GPU rig, Jan is designed to take full advantage of your hardware.

It supports a wide range of platforms:

• NVIDIA GPUs
• Apple M-series
• Intel Macs
• Debian-based Linux
• Windows x64

This flexibility means more people can run cutting-edge models, no data center required.

What You Can Do with Jan

Jan comes packed with features that make local AI practical and powerful:

You can chat with a wide variety of open-source models like Llama, Gemma, Mistral, and Qwen through the built-in Model Library. If you want to connect to external services like Groq or OpenRouter, Jan supports that too, giving you the best of both worlds: offline performance with the option to tap into cloud-based APIs when needed.

There's also a local API server that mimics the OpenAI API, which means developers can build on top of Jan without changing their code.

And yes, Jan even supports chatting with your files, an experimental feature that’s already showing great promise.

Built to Be Customized

Customization is at the core of the Jan experience. With a growing ecosystem of extensions, users can tweak everything from moderation settings to alignment behaviors, or add integrations with their favorite tools.

Coming soon: personalized assistants with long-term memory, allowing Jan to retain context and help automate your workflows over time.

Open, Transparent, and Growing

Jan is developed in the open and shaped by its community. It's fully open-source and already has over 28,000 GitHub stars. You’ll find models hosted on Hugging Face, and an active Discord community where thousands of users share tips, feedback, and enthusiasm.

With over 3 million downloads, Jan isn’t just a promising experiment, it’s a movement.

Take Control of Your AI

Whether you're privacy-conscious, a developer, or just curious about local LLMs, Jan makes advanced AI accessible and empowering. You don’t need to be a machine learning expert to get started, just download Jan, pick a model, and start chatting.

Your data. Your rules. Your AI.

Free. Open Source. Fully Offline.