Jan
Open-source offline ChatGPT alternative from Singapore's Menlo Research
Overview
A free, Apache 2.0 desktop app from Singapore's Menlo Research that runs open-weight LLMs offline on your own machine, with optional connectors to OpenAI, Anthropic, Gemini and others. For developers and privacy-focused users who want a ChatGPT-shaped app without the cloud.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 12 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
What you can produce with Jan
- Open-weight GGUF models run fully offline with a bundled llama.cpp engine
- Models downloaded directly from Hugging Face
- Optional cloud connectors for OpenAI, Anthropic, Gemini, Groq, Mistral, OpenRouter, Azure OpenAI and custom endpoints
- An OpenAI-compatible local API server at 127.0.0.1:1337
- Custom assistants, projects and MCP tool support
- MLX acceleration on Apple Silicon
- Apache 2.0 licensing for commercial use
Jan is what many people mean by ChatGPT on their own laptop. It pulls GGUF models from Hugging Face and runs them offline through a bundled llama.cpp engine (MLX on Apple Silicon), and connectors for OpenAI, Anthropic, Gemini, Groq, Mistral, OpenRouter and others make it a hybrid client. An OpenAI-compatible server on localhost:1337 lets OpenAI-SDK tools use it with a base-URL swap.
It has been Apache 2.0 since May 2025 and releases often. Local quality tracks your hardware, Intel Macs can no longer run local models, and a 2025 Snyk disclosure in its former engine, since fixed, is a reminder that a local server is an attack surface.
What we found
Jan is a free, open-source desktop app for running language models on your own computer, built by Menlo Research Pte. Ltd., a Singapore company with offices in Ho Chi Minh City and Palo Alto. It installs on Windows, Linux and Apple Silicon Macs, downloads GGUF models from Hugging Face and runs them offline through a bundled llama.cpp engine, with MLX available on Macs. Menlo's homepage puts downloads above 6.9 million, its own figure.
The app is a hybrid client as much as a local runner. Its documentation lists connectors for Anthropic, Gemini, Groq, Mistral AI, OpenAI, Azure OpenAI, OpenRouter and Hugging Face, plus custom endpoints, and assistants carry their own instructions. A local server answers OpenAI-style requests at 127.0.0.1:1337, so tools written for the OpenAI SDK can be pointed at it by changing the base URL. Menlo also trains small models for it, including Jan-Nano, Jan-v1 and the March 2026 Jan-Code-4B.
Licensing has moved over the project's life. The code was under AGPLv3 from October 2023 until May 2025, when it moved to Apache 2.0, which suits commercial use; an earlier version of this page described an MIT history that never existed. Releases are frequent: v0.8.6 shipped on 9 October 2026. The v0.8 line dropped local models on Intel Macs and now bundles the inference engine in the installer.
In April 2025 Snyk's research team disclosed flaws in Cortex.cpp, the engine Jan then used: a path traversal that allowed arbitrary file writes, and missing CSRF protection on its local server, which together could let a malicious web page reach the machine. Snyk reported that Menlo acknowledged and fixed the issues quickly.
Menlo's attention is spreading. Its main business is now Asimov, an open-source humanoid robot, and newer Jan work includes a preview command-line agent. Jan itself has no paid tier. For a private, offline chat app with an optional bridge to cloud models, it is a strong free choice, and one built in Singapore.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
About this listing
This entry was compiled from publicly available data including Jan's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Jan unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Jan directly →
Spotted something out of date? Suggest an update →
Related reading
More in LLMs & Chat
