H100 Cloud Rental Pricing
NVIDIA's Hopper flagship with 80GB HBM3, the default for LLM training and high-throughput inference.
21 offers from 12 providers · updated October 4, 2026
Lowest on-demand
$1.68/hr
Median
$3.29/hr
Lowest spot
$1.85/hr
8x GPUs, 1 month
from $9,811
| Provider | Variant | VRAM | On-demand /hr | Spot /hr | Per month | Billing |
|---|---|---|---|---|---|---|
| Latitude.sh | H100 80GB | 80GB | $1.68 | — | $1,226 | per hour |
| Vast.ai | H100 SXM | 80GB | $1.73 | — | $1,263 | per second |
| Vultr | HGX H100 (min 8) | 80GB | $2.30 | — | $1,679 | per hour |
| Vast.ai | H100 PCIe | 80GB | $2.33 | — | $1,701 | per second |
| Hyperstack | H100 PCIe | 80GB | $2.50 | $2.00 | $1,825 | per minute |
| Hyperstack | H100 NVLink | 80GB | $2.60 | — | $1,898 | per minute |
| Vast.ai | H100 NVL | 94GB | $2.60 | — | $1,898 | per second |
| RunPod credits | H100 PCIe | 80GB | $2.89 | — | $2,110 | per ms |
| RunPod credits | H100 NVL | 94GB | $3.19 | — | $2,329 | per ms |
| Hyperstack | H100 SXM | 80GB | $3.20 | — | $2,336 | per minute |
| Lambda Cloud | H100 PCIe | 80GB | $3.29 | — | $2,402 | per hour |
| RunPod credits | H100 SXM | 80GB | $3.49 | — | $2,548 | per ms |
| Oblivus | H100 PCIe | 80GB | $3.50 | — | $2,555 | per hour |
| Oblivus | H100 NVLink | 80GB | $3.64 | — | $2,657 | per hour |
| Verda | H100 SXM | 80GB | $3.70 | $1.85 | $2,701 | per hour |
| Crusoe | H100 HGX | 80GB | $3.90 | — | $2,847 | per minute |
| Lambda Cloud | H100 SXM (min 8) | 80GB | $3.99 | — | $2,913 | per hour |
| Together AI | H100 | 80GB | $3.99 | $1.99 | $2,913 | per hour |
| Oblivus | H100 SXM5 | 80GB | $4.31 | — | $3,146 | per hour |
| Paperspace | H100 | 80GB | $5.95 | — | $4,344 | per hour |
| CoreWeave | HGX H100 (min 8) | 80GB | $6.16 | $2.46 | $4,497 | per hour |
Per GPU-hour, on-demand unless noted. Monthly = 730 hours. Check each provider for region availability and multi-GPU node minimums.
H100 rental FAQ
How much does it cost to rent an H100?
On-demand H100 rental ranges from $1.68 to $6.16 per GPU-hour across 12 providers (median $3.29), as of October 4, 2026. Spot/interruptible capacity starts at $1.85/hr.
What is the cheapest H100 cloud provider?
Latitude.sh has the lowest on-demand H100 price we track at $1.68 per GPU-hour (H100 80GB).
How much does an 8x H100 server cost per month?
At the cheapest on-demand rate ($1.68/GPU-hour), 8x H100 running 24/7 costs about $9,811 per month (730 hours). At the median rate it is about $19,214.