In-browser LLM chat via transformers.js. No server.
Run LFM2.5-1.2B-Thinking directly in your browser on WebGPU