Runs in the browser · Text generation
Run Qwen2.5-0.5B-Instruct in your browser
Qwen2.5-0.5B-Instruct downloads 460.6 MB over WebGPU (q4f16), or 488.4 MB over WebAssembly (uint8), to generate text entirely in the tab with Transformers.js. No install, no server.
Text generation · onnx-community/Qwen2.5-0.5B-Instruct
- Parameters
- 494M
- Pipeline task
- text-generation
Reading
0.93 MB/M params against a 0.73 MB/M text generation median: dense for its parameter count.
The quant ladder spans 4.1x: 460.6 MB (q4f16) to 1.86 GB (fp32).
Run it in your browser
Checking what this browser can run…
All measured variants
| Variant | Size |
|---|---|
| q4f16 WebGPU pick | 460.6 MB |
| uint8 WASM pick | 488.4 MB |
| bnb4 | 728.4 MB |
| q4 | 749.7 MB |
| fp16 | 951.2 MB |
| q8 | 976.7 MB |
| fp32 | 1.86 GB |
Sizes measured from the HuggingFace API file tree for onnx-community/Qwen2.5-0.5B-Instruct, not estimated.
Use it with Transformers.js
import { pipeline } from "@huggingface/transformers";
const pipe = await pipeline("text-generation", "onnx-community/Qwen2.5-0.5B-Instruct", {
device: "webgpu", // usually falls back to "wasm"; wrap in try/catch for production
dtype: "q4f16", // use "uint8" for the WASM build
});
const result = await pipe(/* your input */);
Requires npm install @huggingface/transformers (or the CDN build).
Embed this
A live badge for your model card or README: 460.6 MB over WebGPU, updated as the data is. [](https://localmodel.run/browser/qwen2.5-0.5b-instruct) <a href="https://localmodel.run/browser/qwen2.5-0.5b-instruct"><img src="https://localmodel.run/badge/browser/qwen2.5-0.5b-instruct/size.svg" alt="Qwen2.5-0.5B-Instruct in browser" /></a> Sources
Same job, different size
FAQ
WebGPU or WASM for Qwen2.5-0.5B-Instruct, in practice?
Chrome, Edge, and Safari 26+ run the WebGPU build (q4f16) on the GPU. Firefox has WebGPU on Windows and Apple Silicon Macs but not everywhere yet. Any browser without a WebGPU adapter falls back automatically to the WebAssembly build (uint8) on CPU: same model, slower to load, slower to run.
What does q4f16 mean for Qwen2.5-0.5B-Instruct?
The WebGPU build here uses q4f16: 4-bit weights with some layers kept at 16-bit for stability: WebGPU's usual smallest clean build. The WASM fallback uses uint8: 8-bit integers: about a quarter the size of fp32, with a smaller quality trade than 4-bit.
Sizes measured 2026-08-01 from the HuggingFace API. Last validated 2026-09-14. See all browser models or the methodology.