Jan is a desktop artificial intelligence chat application designed for privacy-conscious users, enabling both offline and online operation. Built to run 100% offline on personal computers and soon on mobile devices, Jan empowers individuals, regardless of coding experience, to download and interact with artificial intelligence models locally. By running all processes on the user´s device, it grants full control over data and ensures conversations remain private. Jan leverages the open-source Llama.cpp backend, providing an API compatible with OpenAI standards, making it possible to connect local third-party tools such as coding assistants or chatbots while using downloaded models or personal cloud-based keys.
The platform supports a wide range of large language models like Llama3, Gemma3, and Qwen3, which can be downloaded directly from the Hugging Face Model Hub or imported in the GGUF format. Beyond local operation, Jan enables integration with cloud artificial intelligence services including OpenAI, Anthropic, Mistral, and Groq when users wish, simply by adding their own API keys in the configuration settings. The intuitive interface facilitates natural conversation with artificial intelligence models, and users can finely tune model parameters for personalized interactions. An API is included to power external apps, mimicking the structure of other well-known artificial intelligence services, further broadening Jan´s flexibility for developers and hobbyists.
Privacy and ownership are at the core of Jan’s philosophy. Distributed under the Apache 2.0 license, Jan is fully open source and self-hostable, encouraging user examination, customization, and community-driven development. All user data remains on the local device, and no information is collected or sold by the creators. The application is compatible with major desktop operating systems — Mac, Windows, and Linux — and supports a variety of GPU architectures for efficiency. Jan’s inspiration draws from the philosophies of calm computing and the local-first movement, seeking to make technology unobtrusive and user-empowering. Community support is robust, with Discord channels and GitHub repositories open for collaboration, contributions, and troubleshooting, reflecting the project´s commitment to openness and user support.
