Apertus Mini WebGPU

Apertus Mini WebGPU

Runs the Apertus Mini model on your own machine inside a web browser, using WebGPU, with no server to set up. It suits developers and curious users who want to try a small model locally.

Made by swiss-ai on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: Apertus Mini WebGPU is a static web page that runs the Apertus Mini model locally in your browser. It uses WebGPU, the browser's graphics interface, so the work happens on your own machine. The Space mirrors a GitHub project where the author takes issues and pull requests.

Who it suits: Developers, researchers and curious users who want to try a small model without installing a server or sending anything to a hosted service. It also suits front-end developers studying in-browser model code.

Title: Who runs models in the browser
Industry | Who uses it | How they use it | Example
Software | Front-end developers | Study how a model runs in a page with WebGPU | Read index.html to see how the page is built
Research | Machine learning researchers | Try a small model locally with no setup | Open the page and test the model on their GPU
Education | Students and teachers | Show a model running on a laptop browser | Demonstrate local inference in class
Hobby | Curious users | Try an open model without an account | Open the page and see it run

If you need a large model or a hosted API for many users, you can skip it.

Where to use it:

• A web browser with WebGPU support: the Space runs on Hugging Face with nothing to install
• Windows, macOS or Linux, by opening the downloaded page in a browser

Use case | Who | What they get back
Try the model with no setup | Anyone | The model running in a browser tab on your own machine
Study in-browser inference | Developers | A small static page that holds the code to read
Demonstrate local models | Teachers | A model that runs without a server

How to set it up: To use it online you need nothing but a browser that supports WebGPU. To keep a copy you need Python and pip for the download.

• A browser with WebGPU support
• Python with pip, only to download a copy

1. Install the Hugging Face hub package and download the Space into a folder.
2. Move into that folder.
3. Load index.html into your browser.

pip install -U huggingface_hub
hf download swiss-ai/apertus-mini-webgpu --repo-type space --local-dir apertus-mini-webgpu
cd apertus-mini-webgpu

How to use it: Open the Space in your browser, or load index.html from your downloaded copy. The page runs the model on your own machine through WebGPU. The source gives no further controls, so use the page as it opens.

cd apertus-mini-webgpu

Example: You open the Space in a browser that supports WebGPU. The page starts the Apertus Mini model on your own graphics hardware, with no remote server doing the work.

For newcomers and experts: If you are new to this, open the Space in a recent desktop browser and let the page load. Everything happens on your machine, so a slower computer may take longer.

If you know the field, this is a model running client side through WebGPU, which moves inference off any server. The trade is that speed and memory depend on the visitor's own hardware and browser. The repository is small, holding an index page, a logo and an assets folder.

You open the page in a browser
The browser uses WebGPU
Apertus Mini runs on your machine

Good to know:

• Browser: it needs a browser with WebGPU support
• Contributing: issues and pull requests go to the GitHub project this Space mirrors
• License: mit
• Cost to run: none named

On Hugging Face, 13 November 2026

• Likes: 12
• Built with: static
• License: mit

Links

• Space: https://huggingface.co/spaces/swiss-ai/apertus-mini-webgpu
• Files: https://huggingface.co/spaces/swiss-ai/apertus-mini-webgpu/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.