What is Jan?
Jan is an open-source ChatGPT replacement for downloading and running large language models with full control and privacy. It runs on Windows, macOS and Linux, and the README says everything runs locally when you want it to.
You can download models such as Llama, Gemma, Qwen and GPT-oss from HuggingFace, or connect to cloud models through OpenAI, Anthropic, Mistral, Groq, MiniMax and others. Jan also lets you create custom assistants for specific tasks.
For developers, Jan exposes an OpenAI-compatible local server at localhost:1337 for other applications and supports the Model Context Protocol for agentic capabilities. It is licensed under Apache 2.0 and is built on llama.cpp and Tauri.
Key capabilities
- Local inference Downloads and runs language models from HuggingFace on the user's computer.
- Cloud connections Connects to providers including OpenAI, Anthropic, Mistral, Groq and MiniMax.
- Custom assistants Creates specialized AI assistants for particular tasks.
- Local API Provides an OpenAI-compatible server for other applications.
- MCP support Integrates Model Context Protocol for agentic capabilities.
- Offline operation Runs local models without relying on cloud inference.
Where Jan fits
Jan fits desktop AI workflows where users want to run models locally, configure task-specific assistants or compare local execution with cloud-provider access. Developers can use its local API to connect applications to models running on their own machine.
Before deployment, users should check operating-system support, memory capacity and GPU compatibility for the intended models. Workflows that require local processing should also verify which model provider is selected, since cloud connections use external services rather than local inference.
Reviews of Jan
Be the first to share your experience with Jan.
Write a review
Loading review form…
