GPUCompare
Real-time cloud GPU pricing — find the best deal
Cheapest H100
Lambda Labs — $2.49/hr
Cheapest A100
Lambda Labs — $1.1/hr
PRAQTOR Edge
Updated every 6 hours from official provider APIs.
See both prices. Spot can save 60-70% for interruptible workloads.
Know which instances are actually available right now.
| Provider | Instance | GPU | Spot | Status | ||
|---|---|---|---|---|---|---|
| GCP | g2-standard-4 | 1x L4 | 24 GB | $0.70/hr | $0.21/hr | Available |
| AWS | g5.xlarge | 1x A10G | 24 GB | $1.01/hr | $0.35/hr | Available |
| Lambda Labs | 1x A100 40GB | 1x A100 | 40 GB | $1.10/hr | — | Limited |
| Lambda Labs | 1x A100 80GB | 1x A100 | 80 GB | $1.29/hr | — | Available |
| RunPod | 1x A100 80GB | 1x A100 | 80 GB | $1.64/hr | $1.34/hr | Available |
| Lambda Labs | 1x H100 | 1x H100 | 80 GB | $2.49/hr | — | Available |
| GCP | a2-highgpu-1g | 1x A100 | 40 GB | $3.67/hr | $1.10/hr | Available |
| Azure | NC24ads_A100_v4 | 1x A100 | 80 GB | $3.67/hr | $1.47/hr | Available |
| RunPod | 1x H100 SXM | 1x H100 | 80 GB | $3.89/hr | $2.69/hr | Available |
| AWS | g5.12xlarge | 4x A10G | 96 GB | $5.67/hr | $1.85/hr | Available |
| Azure | ND96asr_v4 | 8x A100 | 320 GB | $27.20/hr | $8.16/hr | Available |
| AWS | p4d.24xlarge | 8x A100 | 320 GB | $32.77/hr | $10.50/hr | Available |
| AWS | p5.48xlarge | 8x H100 | 640 GB | $98.32/hr | $35.50/hr | Available |
| GCP | a3-highgpu-8g | 8x H100 | 640 GB | $101.22/hr | $30.37/hr | Available |
PRAQTOR Insight
Best value right now: Lambda Labs offers the cheapest H100 at $2.49/hr — 38% cheaper than AWS on-demand. For A100s, Lambda at $1.29/hr beats everyone. If you can tolerate interruptions, GCP spot L4s at $0.21/hr are exceptional for inference.💡 Tip: Reserve capacity for training runs over 24 hours — spot instances will get interrupted. Use spot for development and inference only.
How We Use AI
Data Sources: AWS, GCP, Azure, Lambda Labs, RunPod official pricing pages. Pricing data updated every 6 hours. Spot prices are indicative — check provider dashboards for real-time rates.
Data sources: AWS, GCP, Azure, Lambda Labs, RunPod official pricing pages. Updated every 6 hours.