GPU Rental Prices: Daily Cloud GPU Rental Price Ledger

GPU Rental Prices is a daily, append-only ledger of cloud GPU rental prices across 22 providers, including RunPod, Lambda, CoreWeave, Nebius, Hyperstack, DataCrunch, Voltage Park, Crusoe, Jarvislabs, OVHcloud, Spheron, Microsoft Azure, AWS, Together AI, fal, Modal, Vultr, Massed Compute, TensorDock, SaladCloud, Replicate, and Fireworks AI. Coverage includes on-demand and spot/community pricing for H100, H200, B200, A100, L40S, RTX 4090/5090, MI300X, GH200 and more. Prices are normalized to USD per GPU per hour, excluding storage, egress and networking. Collection method: once per day, each provider's own pricing source is fetched (a public API where one exists, the official pricing page otherwise). Every offer record carries per-offer provenance: the source URL and the exact fetch timestamp. When a fetch fails, that provider's rows are carried forward and flagged with their last-verified date, never silently presented as current. The ledger is append-only, with daily snapshots since 2026-07-05. Contents of this deposit: data/latest.json (most recent snapshot), data/snapshots/YYYY-MM-DD.json (daily snapshots, 2026-07-05 through 2026-07-18), README, LICENSE (CC BY 4.0), CITATION.cff, and METHODOLOGY.txt (plain-text copy of the methodology page). Live resources: Website (sortable tables, history charts, per-cell source and timestamp): https://gpurentalprices.com Live API / data page: https://gpurentalprices.com/data GitHub mirror: https://github.com/adriannutiu/gpu-rental-prices Hugging Face dataset: https://huggingface.co/datasets/gpurentalprices/gpu-rental-prices Methodology: https://gpurentalprices.com/methodology New versions of this record are published quarterly. The concept DOI always resolves to the latest version.