Pick a GPU, choose Data Science, and open JupyterLab with PyTorch preinstalled and your Kaggle and Hugging Face tokens already set up. Prepaid, billed by the minute.
Join the early access groupWelcome to Ubuntu 22.04 computeNexa • 1x RTX 4090 • expires in 3h 59m Your machine: 8 vCPUs · 31 GB RAM JupyterLab is running · What's installed: cnx-info root@cnx-7f3a9c:~# nvidia-smi -L GPU 0: NVIDIA GeForce RTX 4090 root@cnx-7f3a9c:~# kaggle datasets download -d heptapod/titanic Downloading titanic.zip … 100% root@cnx-7f3a9c:~# hf auth whoami alice root@cnx-7f3a9c:~# █
No images to hunt for, no drivers, no copying tokens between machines. Start working on the GPU straight away.
RTX 3090 to H100. Log in with SSH, or open JupyterLab in your browser with one command from the dashboard.
Python, PyTorch, JupyterLab, pandas, scikit-learn and Transformers, plus starter notebooks. See exactly what's included.
Prepaid credits, billed by the minute. Stop early and unused time goes back to your balance.
Save your tokens once. Every GPU you launch has them ready.
kaggle datasets download … just workshf download … just works, including gated models you have access tohf download meta-llama/Llama-3.2-1B
Straight answers to what people ask before they try a new GPU service.
Every GPU is yours alone while it runs; nothing is shared with other customers. When it stops, its disk is wiped.
Kaggle and Hugging Face tokens are checked when you save them, stored encrypted and never shown again. Remove one and it's deleted from your running GPUs within seconds.
Credits are prepaid and billed by the minute from when your GPU is ready. Stop early and unused time goes back to your balance; if a GPU can't start, you get a full refund.
Payments go through Stripe. We never see or store your card number.
Files are deleted when a GPU stops, so download results or push them to Hugging Face or Kaggle before then. Storage that keeps your files between sessions is coming next.
You log in with your own SSH key; there are no passwords to guess. Jupyter is only reachable through your own encrypted connection.
Tell us how you use GPUs. It takes about 90 seconds, and we'll email you when there's a spot.
We invite people in small groups and will email you when your spot is ready. Questions? Email support@computenexa.com.