Jan is an open-source, local-first alternative to ChatGPT that allows users to run AI models entirely offline on their own hardware. Built with a focus on privacy and user control, Jan supports popular open-source LLMs such as Llama 3, Mistral, and Gemma without requiring an internet connection or cloud API subscriptions.
Key Features
- 100% Offline & Private: All conversations, settings, and model data remain locally stored on your device.
- Multi-Model Support: Easily download and run open-source models (Llama 3, Phi-3, Mistral) via an intuitive model hub.
- Cross-Platform: Available for macOS, Windows, and Linux with hardware acceleration (NVIDIA GPU, Apple Silicon, Vulkan).
- API Server & Extensions: Includes a built-in OpenAI-compatible local server and extensible plugin architecture.
Primary Use Cases
- Private AI chatting and document assistance without data privacy concerns.
- Local coding assistance and offline research.
- Developer testing for local LLM integrations via OpenAI-compatible endpoints.
