Load model to device
Runs entirely in your browser. First load downloads ~270 MB then caches locally — instant on next open.
⚠ WebGPU requirements: Chrome 113+ or Edge 113+ on desktop. GPU must support WebGPU.
If unsupported, load will fail — switch back to WASM as fallback.
Qwen2.5-0.5B-Instruct
q4 quantized
~270 MB
WASM · CPU
100% on-device