Runs in the browser · Embeddings

Run voyage-4-nano in your browser

voyage-4-nano downloads 201.3 MB over WebGPU (q4f16), or 402.3 MB over WebAssembly (q8), to turn text into vectors for search and comparison entirely in the tab with Transformers.js. No install, no server.

Embeddings · onnx-community/voyage-4-nano-ONNX

WebGPU download

201.3 MB

q4f16

WASM download

402.3 MB

Parameters: 346.45M
Pipeline task: feature-extraction

Reading

0.58 MB/M params against a 1.03 MB/M embeddings median: light for its parameter count.

The quant ladder spans 6.7x: 201.3 MB (q4f16) to 1.31 GB (fp32).

Live check this browser

Checking for WebGPU support…

All measured variants

Quant Download size

Variant	Size
q4f16 WebGPU pick	201.3 MB
q4	232.0 MB
q8 WASM pick	402.3 MB
fp16	671.0 MB
fp32	1.31 GB

Sizes measured from the HuggingFace API file tree for onnx-community/voyage-4-nano-ONNX, not estimated.

Use it with Transformers.js

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

const pipe = await pipeline("feature-extraction", "onnx-community/voyage-4-nano-ONNX", {
  device: "webgpu", // usually falls back to "wasm"; wrap in try/catch for production
  dtype: "q4f16", // use "q8" for the WASM build
});

const result = await pipe(/* your input */);

Requires npm install @huggingface/transformers (or the CDN build).

Sources

huggingface.co

Same job, different size

← EmbeddingGemma-300M 168.0 MB GTE Multilingual Base → 443.7 MB

FAQ

WebGPU or WASM for voyage-4-nano, 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 (q8) on CPU: same model, slower to load, slower to run.

What does q4f16 mean for voyage-4-nano?

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 q8: 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.