Chat with a local model
The model downloads into your browser and runs on your own device. No server, no API key.
Pick a model
SmolLM2 135M — 112 MB — phones / 3-4 GB RAM
SmolLM2 360M — 260 MB — better answers
Qwen2.5 0.5B — 461 MB — laptop / desktop
Downloaded once, then cached by the browser.
Load model
Send
Runs with
Transformers.js
using WebGPU when available, otherwise CPU.