GPU Economy

GPU prices, per GPU-hour

What it costs to rent a GPU by the hour, and what the same card costs to buy outright. Node-priced offerings are divided by their GPU count, so every figure on this site is directly comparable. Each price links to the provider page it was read from.

Cheapest current rate by model

Lowest tracked per-GPU rate across all providers and price types.
GPUVRAMCheapest / GPU-hourProviderTypeRetail, new
NVIDIA B200192 GB$5.98RunPodon-demand
NVIDIA H200141 GB$3.59RunPodon-demand
NVIDIA H100 SXM80 GB$2.69RunPodon-demand
NVIDIA H100 PCIe80 GB$1.99RunPodon-demand
NVIDIA A100 80GB80 GB$0.56Vast.aion-demand
NVIDIA L40S48 GB$0.53Vast.aion-demand
NVIDIA RTX 409024 GB$0.30Vast.aion-demand
NVIDIA A100 40GB40 GB

Providers

ProviderRegionsModels trackedNotes
RunPod7
Vast.ai7auction pricing, API only
Cudo Compute0
GMI Cloud0
Nebius0
DataCrunch0
Hyperstack0
Jarvislabs0
Lambda0SXM prices are per 8-GPU node - normalize
Modal0per-second serverless
OVHcloud0
Scaleway0
TensorDock0marketplace
Together AI0
Vultr0
CoreWeave08-GPU node prices - normalize; GB200/B300 quote-only
Crusoe0Blackwell quote-only
Latitude.sh0bare metal
Massed Compute0
Paperspace0multi-GPU per-node - normalize
Salad0consumer GPUs only
Voltage Park0
Sesterce0needs headless browser

Data updated: