Train real models in a browser tab, together.

Open a tab and the GPU you already have goes to work: prepare data, run analyses, train models. Nothing to install, no CUDA. Your data stays on your device unless you decide otherwise — and when one machine is not enough, many can share the job.