Run GPT model on the browser with WebGPU. An implementation of GPT inference in less than ~1500 lines of vanilla Javascript.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).