NVIDIA H100 SXM
80 GB HBM3 · Hopper · 700 W
Market Prices — on-demand $/GPU·hr
$2.99 – $12.29/GPU·hr (4.1× spread)
| Provider | GPU | $/GPU·hr | | Observed |
|---|
| RunPod | NVIDIA H100-SXM-80GB | $2.99 | | 1d ago |
| Crusoe | NVIDIA H100-SXM-80GB | $3.90 | | 1d ago |
| Lambda Labs | NVIDIA H100-SXM-80GB | $3.99 | | 1d ago |
| CoreWeave | NVIDIA H100-SXM-80GB | $6.16 | | 1d ago |
| OCI | NVIDIA H100-SXM-80GB | $10.75 | | today |
| GCP (us-central1) | NVIDIA H100-SXM-80GB | $11.06 | | today |
| AWS (us-east-1) | NVIDIA H100-SXM-80GB | $12.29 | | today |
| Azure (eastus2) | NVIDIA H100-SXM-80GB | $12.29 | | today |
| Azure (eastus) | NVIDIA H100-SXM-80GB | $12.29 | | today |
Efficiency by Workload
Offline scenario · on-demand pricing · best result per provider
llama2-70b
| Provider | $/GPU·hr | Perf/GPU | tok/$ | |
|---|
| RunPod | $2.99 | 3,913 tok/s | 4,711,726 tok/$ | |
| Crusoe | $3.90 | 3,913 tok/s | 3,612,323 tok/$ | |
| Lambda Labs | $3.99 | 3,913 tok/s | 3,530,842 tok/$ | |
| CoreWeave | $6.16 | 3,913 tok/s | 2,287,023 tok/$ | |
| OCI | $10.75 | 3,913 tok/s | 1,310,517 tok/$ | |
| GCP (us-central1) | $11.06 | 3,913 tok/s | 1,273,641 tok/$ | |
| AWS (us-east-1) | $12.29 | 3,913 tok/s | 1,146,303 tok/$ | |
| Azure (eastus2) | $12.29 | 3,913 tok/s | 1,146,303 tok/$ | |
| Azure (eastus) | $12.29 | 3,913 tok/s | 1,146,303 tok/$ | |
llama2-70b-interactive
| Provider | $/GPU·hr | Perf/GPU | tok/$ | |
|---|
| RunPod | $2.99 | 3,874 tok/s | 4,663,972 tok/$ | |
| Crusoe | $3.90 | 3,874 tok/s | 3,575,712 tok/$ | |
| Lambda Labs | $3.99 | 3,874 tok/s | 3,495,056 tok/$ | |
| CoreWeave | $6.16 | 3,874 tok/s | 2,263,843 tok/$ | |
| OCI | $10.75 | 3,874 tok/s | 1,297,235 tok/$ | |
| GCP (us-central1) | $11.06 | 3,874 tok/s | 1,260,732 tok/$ | |
| AWS (us-east-1) | $12.29 | 3,874 tok/s | 1,134,685 tok/$ | |
| Azure (eastus2) | $12.29 | 3,874 tok/s | 1,134,685 tok/$ | |
| Azure (eastus) | $12.29 | 3,874 tok/s | 1,134,685 tok/$ | |
mixtral-8x7b
| Provider | $/GPU·hr | Perf/GPU | tok/$ | |
|---|
| RunPod | $2.99 | 6,706 tok/s | 8,074,234 tok/$ | |
| Crusoe | $3.90 | 6,706 tok/s | 6,190,246 tok/$ | |
| Lambda Labs | $3.99 | 6,706 tok/s | 6,050,617 tok/$ | |
| CoreWeave | $6.16 | 6,706 tok/s | 3,919,149 tok/$ | |
| OCI | $10.75 | 6,706 tok/s | 2,245,764 tok/$ | |
| GCP (us-central1) | $11.06 | 6,706 tok/s | 2,182,571 tok/$ | |
| AWS (us-east-1) | $12.29 | 6,706 tok/s | 1,964,358 tok/$ | |
| Azure (eastus2) | $12.29 | 6,706 tok/s | 1,964,358 tok/$ | |
| Azure (eastus) | $12.29 | 6,706 tok/s | 1,964,358 tok/$ | |
resnet50
| Provider | $/GPU·hr | Perf/GPU | samp/$ | |
|---|
| RunPod | $2.99 | 89,080 samp/s | 107,253,512 samp/$ | |
| Crusoe | $3.90 | 89,080 samp/s | 82,227,692 samp/$ | |
| Lambda Labs | $3.99 | 89,080 samp/s | 80,372,932 samp/$ | |
| CoreWeave | $6.16 | 89,080 samp/s | 52,059,740 samp/$ | |
| OCI | $10.75 | 89,080 samp/s | 29,831,442 samp/$ | |
| GCP (us-central1) | $11.06 | 89,080 samp/s | 28,992,022 samp/$ | |
| AWS (us-east-1) | $12.29 | 89,080 samp/s | 26,093,409 samp/$ | |
| Azure (eastus2) | $12.29 | 89,080 samp/s | 26,093,409 samp/$ | |
| Azure (eastus) | $12.29 | 89,080 samp/s | 26,093,409 samp/$ | |
retinanet
| Provider | $/GPU·hr | Perf/GPU | samp/$ | |
|---|
| RunPod | $2.99 | 1,817 samp/s | 2,188,008 samp/$ | |
| Crusoe | $3.90 | 1,817 samp/s | 1,677,473 samp/$ | |
| Lambda Labs | $3.99 | 1,817 samp/s | 1,639,635 samp/$ | |
| CoreWeave | $6.16 | 1,817 samp/s | 1,062,037 samp/$ | |
| OCI | $10.75 | 1,817 samp/s | 608,572 samp/$ | |
| GCP (us-central1) | $11.06 | 1,817 samp/s | 591,447 samp/$ | |
| AWS (us-east-1) | $12.29 | 1,817 samp/s | 532,314 samp/$ | |
| Azure (eastus2) | $12.29 | 1,817 samp/s | 532,314 samp/$ | |
| Azure (eastus) | $12.29 | 1,817 samp/s | 532,314 samp/$ | |
stable-diffusion-xl
| Provider | $/GPU·hr | Perf/GPU | samp/$ | |
|---|
| RunPod | $2.99 | 2 samp/s | 2,503 samp/$ | |
| Crusoe | $3.90 | 2 samp/s | 1,919 samp/$ | |
| Lambda Labs | $3.99 | 2 samp/s | 1,876 samp/$ | |
| CoreWeave | $6.16 | 2 samp/s | 1,215 samp/$ | |
| OCI | $10.75 | 2 samp/s | 696 samp/$ | |
| GCP (us-central1) | $11.06 | 2 samp/s | 677 samp/$ | |
| AWS (us-east-1) | $12.29 | 2 samp/s | 609 samp/$ | |
| Azure (eastus2) | $12.29 | 2 samp/s | 609 samp/$ | |
| Azure (eastus) | $12.29 | 2 samp/s | 609 samp/$ | |
bert
| Provider | $/GPU·hr | Perf/GPU | samp/$ | |
|---|
| RunPod | $2.99 | 9,110 samp/s | 10,967,960 samp/$ | |
| Crusoe | $3.90 | 9,110 samp/s | 8,408,769 samp/$ | |
| Lambda Labs | $3.99 | 9,110 samp/s | 8,219,098 samp/$ | |
| CoreWeave | $6.16 | 9,110 samp/s | 5,323,734 samp/$ | |
| OCI | $10.75 | 9,110 samp/s | 3,050,623 samp/$ | |
| GCP (us-central1) | $11.06 | 9,110 samp/s | 2,964,782 samp/$ | |
| AWS (us-east-1) | $12.29 | 9,110 samp/s | 2,668,365 samp/$ | |
| Azure (eastus2) | $12.29 | 9,110 samp/s | 2,668,365 samp/$ | |
| Azure (eastus) | $12.29 | 9,110 samp/s | 2,668,365 samp/$ | |
dlrm-v2
| Provider | $/GPU·hr | Perf/GPU | samp/$ | |
|---|
| RunPod | $2.99 | 75,264 samp/s | 90,618,261 samp/$ | |
| Crusoe | $3.90 | 75,264 samp/s | 69,474,000 samp/$ | |
| Lambda Labs | $3.99 | 75,264 samp/s | 67,906,917 samp/$ | |
| CoreWeave | $6.16 | 75,264 samp/s | 43,985,162 samp/$ | |
| OCI | $10.75 | 75,264 samp/s | 25,204,521 samp/$ | |
| GCP (us-central1) | $11.06 | 75,264 samp/s | 24,495,297 samp/$ | |
| AWS (us-east-1) | $12.29 | 75,264 samp/s | 22,046,265 samp/$ | |
| Azure (eastus2) | $12.29 | 75,264 samp/s | 22,046,265 samp/$ | |
| Azure (eastus) | $12.29 | 75,264 samp/s | 22,046,265 samp/$ | |
3d-unet
| Provider | $/GPU·hr | Perf/GPU | samp/$ | |
|---|
| RunPod | $2.99 | 7 samp/s | 7,871 samp/$ | |
| Crusoe | $3.90 | 7 samp/s | 6,034 samp/$ | |
| Lambda Labs | $3.99 | 7 samp/s | 5,898 samp/$ | |
| CoreWeave | $6.16 | 7 samp/s | 3,820 samp/$ | |
| OCI | $10.75 | 7 samp/s | 2,189 samp/$ | |
| GCP (us-central1) | $11.06 | 7 samp/s | 2,128 samp/$ | |
| AWS (us-east-1) | $12.29 | 7 samp/s | 1,915 samp/$ | |
| Azure (eastus2) | $12.29 | 7 samp/s | 1,915 samp/$ | |
| Azure (eastus) | $12.29 | 7 samp/s | 1,915 samp/$ | |