Jan

Jan — Open source offline ChatGPT-alternative

Jan is an open source AI assistant that runs 100% offline on your device. It emphasizes local-first design, user ownership of data, and full customization. You can chat with AI locally, load and run powerful models on your computer, and optionally connect to cloud-based AIs when needed.

How it works

• Local AI: Run conversations with AI models directly on your machine for privacy-conscious interactions.
• Model Hub: Download and run models like Llama3, Gemma, or Mistral locally.
• Cloud Connectors: Route requests to more capable cloud models (e.g., OpenAI, Groq, Cohere) when desired.
• Local API Server: Set up an OpenAI-compatible API server powered by local models with a single click.
• Extensible: Extend Jan with 3rd party extensions, data connectors, and cloud AI connectors.

Key principles

• Local-first: Conversations and files stay on your device; privacy and ownership are prioritized.
• User-owned: Your data remains under your control and portable across systems.
• Fully customizable: Tailor alignment, moderation, and censorship levels; install extensions to fit your workflow.

Core strengths

• Offline operation: Use Jan without an internet connection.
• Open source: Transparent development with public repositories and community contributions.
• Cross-ecosystem support: Deploy local models or connect to cloud AIs as needed.
• Active development: Ongoing updates, models, and features (e.g., memory/assistance features coming soon).

How to use Jan

1. Download and install on macOS (or other supported platforms) from the official repository.
2. Choose your mode: run local models, or connect to cloud AIs for enhanced capabilities.
3. Start chatting with the AI, or set up a local API server for programmatic access.
4. Customize with extensions and alignment/moderation settings to fit your needs.

Note: Jan is designed to be highly configurable and privacy-respecting, with all critical data processed locally unless you opt-in to cloud services.

Feature list

• Local-first AI: Conversations and data stay on your device by default
• 100% offline operation: Run models locally without requiring internet access
• Model Hub: Download and run models like Llama3, Gemma, Mistral locally
• Cloud AIs integration: Route to OpenAI, Groq, Cohere, and other cloud providers when needed
• Local API server: One-click setup for an OpenAI-compatible API
• Extensions: Extend Jan with 3rd party connectors, tools, and data integrations
• Highly customizable: Adjust alignment, moderation, and censorship levels
• Open Source: Transparent development with public repositories and community involvement
• Memory and assistants (coming soon): Create personalized assistants that remember conversations and tasks

Safety and privacy considerations

• Local-first design prioritizes privacy; data remains on your device by default.
• You can opt to connect to cloud providers if you need capabilities beyond local models.
• As with all AI tools, manage how and where you deploy AI to align with your privacy and security requirements.

Technical details

• Platform: macOS download available; ongoing cross-platform support
• Licensing: Free & Open Source
• Community: Active ecosystem with multiple contributors and ongoing updates

---

Quick start (summary)

• Download Jan for your platform
• Run local models or connect to cloud AIs
• Optional: set up Local API server for programmatic use
• Customize with extensions and settings to suit your workflow