Free tool
In-Browser LLM Sandbox
Run a small open-weights model entirely on your own machine — no API key, no cost, and nothing you type ever leaves your browser. The model downloads once to your device (uses your GPU via WebGPU), not our servers. Perfect for safely poking at how open models behave.
Runs on your device via WebGPU. The model downloads once from the open MLC/Hugging Face CDN to your browser cache — not our servers — and nothing you type is uploaded. First load is a sizeable download; after that it’s instant and offline-capable.
Load a model to start chatting.