In Operation
Here’s Jan in action. We’ve asked the tool a simple programming question.
The result was generated with Llama 3.2 1B Instruct Q8. The software makes it very easy to download and install a whole raft of other models without even knowing the Hugging Face URL.
We like that the interface flags any models which will slow on your device, and models where your system has insufficient RAM.
The interface also lets you access a variety of models remotely including from Anthropic, Cohere, Deepseek, Google, Groq, Martian, Mistral, NVIDIA, OpenAI, and OpenRouter.
Other features:
- Local API server – setup and run your own OpenAI-compatible API server using local models.
- Chat with your files.
- Customize Jan with extensions.
Summary
Jan is an excellent open source alternative to OpenAI. Run it on your computer or server locally. We particularly like the interface which is well designed and easy to use. Naturally, the quality of the output generated from the models is out of Jan’s control.
The project has amassed over 27k GitHub stars.
Jan is powered by Cortex.cpp. That’s a C++ command-line interface (CLI) designed as an alternative to Ollama. By default, it runs on the llama.cpp engine but also supports other engines, including ONNX and TensorRT-LLM, making it a multi-engine platform.
Website: github.com/janhq/jan
Support:
Developer: Homebrew Computer Company
License: GNU Affero General Public License v3.0
For other useful open source apps that use machine learning/deep learning, we’ve compiled this roundup.
Jan is written in TypeScript. Learn TypeScript with our recommended free books and free tutorials.
Pages in this article:
Page 1 – Introduction and Installation
Page 2 – In Operation and Summary