Normalized published cloud GPU rates across providers — an on-demand index plus separately labeled reservation and marketplace rows, with citable sources and real-data history.
Start with live cloud rates, then add your hardware, power, and operating assumptions.
The headline index median uses comparable cloud GPU listings and reports prices in USD per GPU per hour, so H100, H200, B200, A100, L40S, RTX 4090, and RTX 5090 pages stay consistent.
Live price tables show provider, billing mode, quality tier, source URL, and normalization method. Every number keeps enough context to verify the source listing.
Trend charts use recorded snapshots and archive backfills only. Missing history remains missing rather than being interpolated into a fake GPU price graph.