A real LLM on your own GPU. No account, no API key, no server — ask it anything on a plane, in a village with no signal, or with secrets you'd never paste into ChatGPT. Nothing leaves this tab.
Downloads once, cached forever · runs 100% on your GPU
The language model itself downloads into your browser and runs on your GPU. Your messages are never sent to a server — there is no server. Close the tab and the conversation is gone unless you save it.
Yes — once the model has downloaded, you can disconnect entirely and keep chatting. Useful on flights, and proof that nothing is leaving your machine.
Chrome, Edge or Brave on a machine with a GPU — the model engine needs full WebGPU. Safari can't run this one yet; that's a browser limitation, not a signup wall.
Genuinely useful for drafting, summarising, coding help and questions — not at the level of the biggest cloud models. The trade is absolute privacy and zero cost, and for most day-to-day prompts it holds up well.