Assistant · free, in your browser

Can my PC run a local LLM?

Assistant reads what your browser reports about your GPU, runs a short WebGPU benchmark, and tells you for each model whether it runs well, runs slowly or will not fit.

Check my PC

How to do it

1. Open Assistant from this page: it starts on Can my PC run it?.
2. Read what it finds: WebGPU support, the GPU adapter, its memory limits and shader-f16, RAM, CPU threads, browser and storage.
3. Let the benchmark run, about one second of GPU work, for matrix speed in GFLOPS and memory bandwidth in GB/s. Run again repeats it.
4. See the verdict for each model, with an estimated speed or the reason it is slow or will not fit. Get takes you to that model.
5. Press Spec card to download your results as a picture, save them to Files or share them.

Your files stay on this device: Assistant runs in your browser, with no upload, no account and no watermark.

Questions

What if WebGPU is missing?

The page says so and shows how to turn it on in your browser. You can also try the demo engine, which gives canned replies without downloading a model.

Why does it say 8 GB of RAM when I have more?

Browsers report at most 8 GB of memory. For these models the GPU limits matter more, and those are read from WebGPU directly.

Does the check download anything?

No. The check and the benchmark use your GPU for about a second and download nothing. A model downloads only when you choose one and confirm.

How good are the speed estimates?

They come from the memory bandwidth benchmark, so they are a guide. The real speed shows under every reply once you chat.

More with Assistant

• Run a local AI model in your browser
• A private, free ChatGPT alternative that runs on your computer

Related apps

• Compare
  Beta
  Compare two texts, files, PDFs or pictures and see every change. Nothing uploaded.
• PDF Editor
  Available
  Organise, edit and sign PDFs. Nothing uploaded.
• Label Designer
  Beta
  Design and print barcode labels.