What is this?
Jan is an open-source, self-hosted alternative to ChatGPT that runs 100% offline on your own computer, letting you download and run LLMs locally with full control and privacy.
Jan is an open-source, self-hosted alternative to ChatGPT that runs 100% offline on your own computer, letting you download and run LLMs locally with full control and privacy.Is this a desktop app or something I run in a terminal?It's a desktop application built with Tauri (Rust backend), with installers for Windows, macOS, and Linux (deb/AppImage/Arm64).
It's a desktop application built with Tauri (Rust backend), with installers for Windows, macOS, and Linux (deb/AppImage/Arm64).How do I get it running?Download the prebuilt installer for your OS from jan.ai or GitHub Releases, or build it from source with Node.js, Yarn, Make, and Rust using 'make dev'.
Download the prebuilt installer for your OS from jan.ai or GitHub Releases, or build it from source with Node.js, Yarn, Make, and Rust using 'make dev'.Any gotchas when building on Windows?You must run 'make dev' from Git Bash, not cmd.exe, since make dispatches recipes through sh; also long build paths can hit Windows' 260-character MAXPATH limit, which the build script tries to work around automatically.
You must run 'make dev' from Git Bash, not cmd.exe, since make dispatches recipes through sh; also long build paths can hit Windows' 260-character MAXPATH limit, which the build script tries to work around automatically.What hardware do I need to run bigger models smoothly?Minimum specs scale with model size: 8GB RAM for 3B models, 16GB for 7B, and 32GB for 13B models; GPU acceleration is available on Windows (NVIDIA/AMD/Intel Arc) and Linux.
Minimum specs scale with model size: 8GB RAM for 3B models, 16GB for 7B, and 32GB for 13B models; GPU acceleration is available on Windows (NVIDIA/AMD/Intel Arc) and Linux.Does my data leave my machine?No — everything runs locally by default, and cloud connections (OpenAI, Anthropic, etc.) are only used if you explicitly set them up.
No — everything runs locally by default, and cloud connections (OpenAI, Anthropic, etc.) are only used if you explicitly set them up.Can Jan connect to cloud AI providers too, or is it strictly local?It can do both: it runs local models fully offline, but also supports Cloud Integration to connect to GPT via OpenAI, Claude via Anthropic, Mistral, Groq, and MiniMax.
It can do both: it runs local models fully offline, but also supports Cloud Integration to connect to GPT via OpenAI, Claude via Anthropic, Mistral, Groq, and MiniMax.Does Jan expose an API for other apps to use?Yes, it runs an OpenAI-Compatible API on a local server at localhost:1337 so other applications can talk to it.
Yes, it runs an OpenAI-Compatible API on a local server at localhost:1337 so other applications can talk to it.Can Jan do more than chat, like use tools or agents?Yes, it has Model Context Protocol (MCP) integration for agentic capabilities and lets you create Custom Assistants for specialized tasks.
Yes, it has Model Context Protocol (MCP) integration for agentic capabilities and lets you create Custom Assistants for specialized tasks.Where do the local models come from?You download and run LLMs such as Llama, Gemma, Qwen, and GPT-oss directly from HuggingFace.
You download and run LLMs such as Llama, Gemma, Qwen, and GPT-oss directly from HuggingFace.What are similar projects?llama-gpt, unsloth, Bonsai-demo, ThinkWatch-Lite, gpt4free, laya, ruflo, NextChat
llama-gpt, unsloth, Bonsai-demo, ThinkWatch-Lite, gpt4free, laya, ruflo, NextChatWhich topics is it in?localai, llamacpp, gguf, tauri, gpt, chatgpt, openai-compatible, local-llm, model-context-protocol, chatbot, inference, huggingface
localai, llamacpp, gguf, tauri, gpt, chatgpt, openai-compatible, local-llm, model-context-protocol, chatbot, inference, huggingface