You are offline — chat still works if the model is already loaded
✦
Local AI Chat
WebGPU … No model
Preparing…

Private AI on your device

Powered by WebLLM + WebGPU. Models run entirely in your browser — nothing is sent to a server.

1. Choose a model & click Load model
2. First load downloads weights (cached afterwards)
3. Chat offline with GPU acceleration

Load a model to start 100% local · WebGPU accelerated