Run real AI models in your browser, on your own GPU

This is a WebGPU playground for open-source AI. Pick a task, pick a model from the Hugging Face ONNX catalogue, and it downloads straight into your browser cache and runs on your graphics card via WebGPU — chat with a small language model, transcribe audio with Whisper, cut backgrounds out of photos, detect objects, embed text, translate, summarise. There is no account, no API key and no server-side inference: after the first download, everything happens on your machine and works with the network switched off.

browser aiwebgpu airun ai in browsertransformers.js playgroundlocal ai no installhuggingface in browserclient side machine learningprivate ai toolsoffline aionnx runtime webfree ai without api keyon device inference

Frequently asked questions

What do I need to run these models?

A browser with WebGPU — Chrome or Edge 113+, or Safari 26+. A discrete GPU makes larger models usable, but small models (35–150 MB) run fine on integrated graphics, and there is a WebAssembly fallback for browsers with no WebGPU at all.

Is my data uploaded to a server?

No. The model weights are downloaded from Hugging Face to your browser once, and every inference after that runs on your own GPU through WebGPU. Your text, images and audio are never sent anywhere — hjLabs.in has no inference server and no way to see your input.

Is it really free, and is there a usage limit?

It is free and there is no per-request limit, because we are not paying for the compute — your own GPU is. The only cost is the one-time model download, which your browser caches.

How big are the downloads?

From about 25 MB for a sentiment classifier or a MiniLM embedding model, up to roughly 1–2 GB for a 1.7B-parameter chat model. Each model page shows the exact size before you commit to it.

Explore the rest of the playground

Text & language models

LLM chat — Chat with a local LLM running on your own GPU
Text embeddings — Generate text embeddings locally, in the browser
Translation — Translate text offline with a local neural MT model
Summarization — Summarise long text with a model running on your machine
Sentiment analysis — Sentiment analysis that runs in your browser
Zero-shot text classification — Classify text into labels you make up on the spot
Named entity recognition — Named entity recognition, running on your own machine
Question answering — Ask questions about a passage, get the exact answer span
Fill-mask — Watch a masked language model guess the missing word

Audio models

Speech to text (Whisper) — Transcribe audio with Whisper, entirely on your device
Text to speech (Kokoro) — Turn text into natural speech, locally
Audio classification — Classify what a sound actually is, in the browser

Vision models

Background removal — Remove a background without uploading the photo
Object detection — Detect objects in an image, on your own hardware
Zero-shot object detection — Detect anything you can describe, with no training
Image classification — Classify an image with ViT, locally
Zero-shot image classification — Classify images against labels you invent, using CLIP
Image segmentation — Segment an image into labelled regions, locally
Depth estimation — Generate a depth map from a single photo
Image to 3D (point cloud) — Turn a single photo into a 3D scene you can orbit
Image captioning — Generate a caption or alt text for any image
OCR (image to text) — Extract text from images without uploading them

Guides & comparisons

Model catalog — The model catalog: what actually runs in a browser
WebGPU vs WASM — WebGPU vs WebAssembly for running models in a browser
Browser AI vs cloud API — Browser inference vs a hosted API: an honest comparison
Transformers.js vs WebLLM — Transformers.js vs WebLLM
Whisper model sizes — Choosing a Whisper size for browser transcription
fp32 vs q8 vs q4 — Picking a quantization level for browser inference
Why no image generation — Why this playground has no text-to-image generation
FAQ — Frequently asked questions about browser-based AI
About — About this project

This playground needs JavaScript and WebGPU. Models are downloaded to your browser and executed on your own GPU, so there is nothing to show without a script engine. Enable JavaScript, or read the FAQ for what the site does.