Skip to content

Runs in the browser · Vision

Run Depth Anything V2 Small in your browser

Depth Anything V2 Small downloads 18.2 MB over WebGPU (q4f16), or 26.0 MB over WebAssembly (uint8), to estimate depth from a single image entirely in the tab with Transformers.js. No install, no server.

Vision · onnx-community/depth-anything-v2-small

WebGPU download
18.2 MB
q4f16
WASM download
26.0 MB
uint8
Parameters
24.8M
Pipeline task
depth-estimation

Reading

0.74 MB/M params against a 1.73 MB/M vision median: light for its parameter count.

The quant ladder spans 5.2x: 18.2 MB (q4f16) to 94.5 MB (fp32).

Will it run in your browser?

Live check this browser

Checking for WebGPU support…

All measured variants

Quant Download size
Variant Size
q4f16 WebGPU pick 18.2 MB
bnb4 24.9 MB
uint8 WASM pick 26.0 MB
q4 26.1 MB
fp16 47.3 MB
q8 52.0 MB
fp32 94.5 MB

Sizes measured from the HuggingFace API file tree for onnx-community/depth-anything-v2-small, not estimated.

Use it with Transformers.js

import { pipeline } from "@huggingface/transformers";

const pipe = await pipeline("depth-estimation", "onnx-community/depth-anything-v2-small", {
  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).

Sources

Same job, different size

FAQ

WebGPU or WASM for Depth Anything V2 Small, 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 Depth Anything V2 Small?

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 2026-08-03. See all browser models or the methodology.