Jan for MAC

Free, open-source AI app for Mac that downloads and runs large language models locally, with optional cloud models, assistants and MCP.

Jan Overview

Jan is a free, open-source app for running AI models on your own Mac. You can download large language models such as Llama, Gemma, Qwen and GPT-oss from Hugging Face and chat with them locally, so everything can stay on your computer. When you want, Jan can also connect to cloud models from OpenAI, Anthropic, Mistral, Groq, MiniMax and others with your own API keys. Custom assistants, a local OpenAI-compatible API server on localhost:1337 and Model Context Protocol support make it useful for agent and developer work too.

Jan is made by Menlo Research and released under the Apache 2.0 licence. Version 0.8.8 is a security update that the developer recommends installing; it also updates the bundled llama.cpp engine and adds Trusted Hosts for the local API. The Mac app is universal, for Apple Silicon and Intel Macs. The app itself starts on macOS 10.13 or later, but for a decent experience Jan recommends macOS 13.6 or later with 8 GB of memory for 3B models, 16 GB for 7B and 32 GB for 13B models.

Key Features

Jan brings open-source AI models to your Mac with full control and privacy.

  • Local models: download and run LLMs from Hugging Face on your Mac.
  • Offline use: local models work without sending your chats anywhere.
  • Cloud models: connect OpenAI, Anthropic, Mistral, Groq and others.
  • Custom assistants: create assistants for your own tasks.
  • Local API: an OpenAI-compatible server at localhost:1337.
  • MCP support: Model Context Protocol tools for agentic tasks.
  • Trusted Hosts: control which web pages may call the local API.
  • Cross-platform: also on Windows and Linux.

What’s New in Version 0.8.8

Jan 0.8.8 is a security update with llama.cpp 0.7.0 and Trusted Hosts for the local API.

  • Links in chats, tool results and model cards always open in your system browser
  • The local API server only accepts browser requests from trusted origins
  • Server-side tool execution stays opt-in for every endpoint
  • File paths are checked so threads and extensions stay inside Jan's data folder
  • The Claude Code integration writes your shell profile more safely

Jan FAQ

Is Jan free?

Yes. Jan is free and open source under the Apache 2.0 licence. Cloud models you connect use your own API keys and may cost money.

Can Jan run AI models offline?

Yes. You can download models from Hugging Face and run them locally on your Mac, without an internet connection.

Which models can I use in Jan?

Local models such as Llama, Gemma, Qwen and GPT-oss, and cloud models from OpenAI, Anthropic, Mistral, Groq, MiniMax and others.

How much memory does Jan need?

Jan recommends 8 GB of memory for 3B models, 16 GB for 7B models and 32 GB for 13B models.

Which macOS version does Jan need?

The app starts on macOS 10.13 or later, but Jan recommends macOS 13.6 or later for a decent experience.

Does Jan run on Apple Silicon?

Yes. The Mac download is a universal app for Apple Silicon and Intel Macs.

Does Jan have an API?

Yes. Jan can run a local OpenAI-compatible API server at localhost:1337 for other apps.

Does Jan support MCP?

Yes. Jan supports the Model Context Protocol for agentic tools. Server-side tool execution for the local API is opt-in.

Should I update to Jan 0.8.8?

Yes. The developer calls 0.8.8 a security update and recommends updating; your data and settings carry over.

Where does the Jan download on jan.FULLMAC.app come from?

The download button leads to the official Jan download page at https://www.jan.ai/download. jan.FULLMAC.app does not change or repackage the app.

Similar Apps

FULLMAC.app does not host this app. Download buttons lead to the developer’s own pages.