Atomic Chat

Atomic Chat is an open-source desktop application that serves as a privacy-first alternative to ChatGPT, built with Tauri and TypeScript. It lets you download and run large language models locally on your machine or connect to cloud-based AI providers like OpenAI, Anthropic, and Mistral. All local inference runs entirely offline, giving you complete control over your data. The app also exposes an OpenAI-compatible local API server and supports the Model Context Protocol for agentic workflows.

• Local AI Models — Download and run LLMs (Llama, Gemma, Qwen, and more) directly from HuggingFace without any cloud dependency.
• Cloud Integration — Connect to OpenAI, Anthropic, Mistral, Groq, MiniMax, and other cloud providers from a single interface.
• Custom Assistants — Create specialized AI assistants tailored to specific tasks or workflows.
• OpenAI-Compatible API — A local server runs at localhost:1337, making Atomic Chat a drop-in backend for other applications.
• Model Context Protocol (MCP) — Built-in MCP integration enables agentic capabilities and tool use.
• Privacy First — All processing happens locally when using local models; no data leaves your machine.
• Built with Tauri — Lightweight, cross-platform desktop app built on Rust and web technologies for a native feel.
• Open Source (Apache 2.0) — Full source code available on GitHub; contributions and forks are welcome.

To get started, download the macOS .dmg from the GitHub Releases page or the official website, install the app, and either pull a model from HuggingFace or enter your API key for a cloud provider. Developers can build from source using Node.js ≥ 20, Yarn ≥ 4.5.3, and Rust (for Tauri) by running make dev after cloning the repository.