Qwen2.5‑Coder‑3B‑Instruct

First run will download ~2 GB+ (q4* ONNX weights). Desktop Chrome/Edge recommended. HTTPS or localhost required for WebGPU.

Model ID or local path

If you mirror locally, ensure it contains an onnx/ directory.

Device

dtype (quantization/precision)

Local mirror (optional)

Local only (disable remote)

Example mirror path: /models/onnx-community/Qwen2.5-Coder-3B-Instruct/onnx/*

This will download model files (potentially several GB) and run inference via WebGPU. Proceed?

Auto‑run on click

Qwen2.5‑Coder‑3B‑Instruct — WebGPU (ONNX)