A live price index for GPU cloud compute.
Every tracked GPU, every cloud, one comparable number — refreshed every six hours from open data.
What NeoScalers is
NeoScalers is an independent, live price index for GPU cloud compute. It tracks on-demand per-GPU rental prices across cloud providers — from the NVIDIA H100 and B200 to the RTX 4090 and AMD Instinct MI300X — and re-ranks them roughly every six hours, so buyers can see who charges what for the same silicon in seconds instead of reading a dozen pricing pages.
Where the numbers come from
Every price on this site is sampled from two open sources:
- The SkyPilot catalog — open source (Apache-2.0), a machine-readable catalog of cloud instance prices.
- The gpurentalprices.com feed — a licensed price feed (CC BY 4.0).
Each sample is timestamped; every table and page carries its sample date. Prices are normalized to per-GPU-hour (node price ÷ GPU count) so different instance shapes stay comparable, and failed fetch runs never become the served snapshot — the last good sample stays up.
Neutrality policy
Ranking is by price alone. NeoScalers shows no provider logos, sells no rankings, and runs no paid placement. No financial relationship with any listed provider affects ordering. If a provider offered money for a better position, the answer would be no.
What NeoScalers is not
Not a cloud provider. Not a broker or agent of any provider. The prices shown are informational on-demand list-price samples — availability, regions, and terms change constantly. Always verify current pricing and capacity with the provider before buying.
The quote service
Separate from the index, the quote service takes your GPU specs and has them reviewed by a person, who then connects you with providers that fit. The index itself stays fully automated and fully neutral.
Who it's for
AI teams, researchers, and startups buying GPU compute — anyone who would rather compare numbers than read marketing pages.