GPU.ai is active here. We do not have a founding year or a headquarters on file. Billing is per hour. The recorded payment method is credit card. The published catalogue is long: H100 SXM, H100 PCIe, H200 SXM, B200, A100, L40, L40S, L4, A40, RTX 4090, RTX 5090, RTX 6000 Ada, RTX A6000, and RTX PRO 6000.
Why GPU.ai stands out
Our data records multi-GPU, an API, Docker, and spot instances. Persistent storage, Kubernetes, Jupyter, reserved instances, and SOC 2 are not set. There is a public hourly list. The old line about no inventory does not match the card.
Pros
- Many published NVIDIA rows, including H100, H200, B200, and 4090
- Per-hour billing and spot recorded
- API and Docker recorded
Cons
- No Jupyter, no Kubernetes, and no persistent storage in our data
- No reserved instances and no SOC 2 flag
- No founding details on file
- Enterprise readiness in our scoring is low
Getting started
- Open GPU.ai and compare a spot or on-demand row with this page.
- Pay by credit card.
- Plan on Docker or the API. Jupyter and Kubernetes are not recorded.
Best for: An hourly or spot NVIDIA card, including H100 and 4090, when you can bring your own notebook environment.