REDACTED1/llm-chat-demo

Qwen2.5-1.5B-Instruct — in your browser, free

A chat demo powered by transformers.js running the model entirely in your browser via WebGPU — nothing is sent to a server.

• —Model: onnx-community/Qwen2.5-1.5B-Instruct (q4f16, ~1.2 GB, cached after first load)
• —Requires a WebGPU-capable browser (Chrome/Edge 113+, recent Firefox/Safari)
• —Falls back to WASM (slower) when WebGPU is unavailable

First load downloads ~1.2 GB of weights, then everything runs locally.