LIVE — Updated every 30 min
AWS GPU Spot Prices —
Real-Time AWS GPU Pricing
Live spot and on-demand GPU prices on Amazon Web Services (AWS). Amazon EC2 GPU instances including P4 (A100) and G5 (A10G) families. Updated from live provider APIs every 30 minutes.
8
GPUs Available
$0.10
Cheapest Spot /hr
8
GPUs w/ Spot Pricing
| GPU Model | Spot Price | On-Demand | Savings | Availability |
|---|---|---|---|---|
| NVIDIA H100 SXM80GB | $31.30/hr | $98.32/hr | 68% | Spot Available |
| NVIDIA A100 40GB40GB | $8.10/hr | $19.22/hr | 58% | Spot Available |
| NVIDIA A100 SXM80GB | $8.70/hr | $32.77/hr | 73% | Spot Available |
| NVIDIA L40S48GB | $1.82/hr | $3.22/hr | 43% | Spot Available |
| NVIDIA A10G24GB | $0.33/hr | $1.62/hr | 79% | Spot Available |
| NVIDIA L424GB | $0.10/hr | $0.81/hr | 87% | Spot Available |
| NVIDIA T416GB | $0.35/hr | $0.75/hr | 54% | Spot Available |
| NVIDIA V10016GB | $9.96/hr | $24.48/hr | 59% | Spot Available |
About AWS GPU Pricing
Amazon EC2 GPU instances including P4 (A100) and G5 (A10G) families. Spot instances use unused capacity at 60–90% discounts vs on-demand pricing. The tradeoff: providers can reclaim instances with short notice (typically 30 seconds). For fault-tolerant workloads like batch inference, training jobs, and data pipelines, spot pricing cuts GPU costs dramatically.
RoofRun monitors AWS GPU prices every 30 minutes alongside 6 other providers, so you can compare and switch when pricing shifts. Compare all providers →