H100 · 1-GPU configurations
Read 12 November 2026
One H100: the 1-GPU price per hour
- Lowest per GPU
- $1.99/GPU-hr
- Voltage Park
- One day, one GPU
- $47.76
- 24 hours at $1.99
- Median per GPU
- $2.99/GPU-hr
- of 13 providers
- Highest per GPU
- $10.00/GPU-hr
- Oracle Cloud
Basis: 21 current on-demand rows with exactly 1 GPU, one lowest row kept per provider. Marketplace figures and announced prices are shown but not counted. A day is 24 hours and a month is 720 hours.
Every 1-GPU listing
29 current rows| Provider | Configuration | Billing | $/GPU-hr | $/hr, config | Region | Read | Source |
|---|---|---|---|---|---|---|---|
| RunPodmarketplace | 1× PCIeCommunity Cloud H100 PCIe | per second | $1.99 | $1.99 | not stated | 2026-11-12 | page |
| Voltage Parkstarting price | 1×On-demand H100 (FAQ starting price) | per hour | $1.99 | $1.99 | not stated | 2026-11-12 | page |
| Vast.aimarketplace | 1× PCIeMarketplace lowest listed (H100 PCIE) | not stated | $2.00 | $2.00 | not stated | 2026-11-12 | page |
| TensorDockmarketplace | 1× SXMMarketplace from-price (H100 SXM5) | not stated | $2.25 | $2.25 | not stated | 2026-11-12 | page |
| Hyperstack | 1× PCIeNVIDIA H100 | per minute | $2.50 | $2.50 | not stated | 2026-11-12 | page |
| Vast.aimarketplace | 1× PCIeMarketplace median (H100 PCIE) | not stated | $2.53 | $2.53 | not stated | 2026-11-12 | page |
| Hyperstack | 1×NVIDIA H100 NVLink (NVLink) | per minute | $2.60 | $2.60 | not stated | 2026-11-12 | page |
| Vast.aimarketplace | 1× SXMMarketplace lowest listed (H100 SXM) | not stated | $2.67 | $2.67 | not stated | 2026-11-12 | page |
| RunPodmarketplace | 1× SXMCommunity Cloud H100 SXM | per second | $2.69 | $2.69 | not stated | 2026-11-12 | page |
| Massed Compute | 1×H100 (80GB) | not stated | $2.73 | $2.73 | not stated | 2026-11-12 | page |
| Vast.aimarketplace | 1× NVLMarketplace lowest listed (H100 NVL) | not stated | $2.80 | $2.80 | not stated | 2026-11-12 | page |
| RunPod | 1× PCIeSecure Cloud H100 PCIe | per second | $2.89 | $2.89 | not stated | 2026-11-12 | page |
| Thunder Compute | 1× SXMSandbox H100 SXM | per second | $2.95 | $2.95 | not stated | 2026-11-12 | page |
| Civo | 1× SXMSmall 1 x NVIDIA H100 - 80GB | per hour | $2.99 | $2.99 | not stated | 2026-11-12 | page |
| OVHcloud | 1×h100-380 | per hour | $2.99 | $2.99 | not stated | 2026-11-12 | page |
| Massed Compute | 1× NVLH100 NVL | not stated | $3.11 | $3.11 | not stated | 2026-11-12 | page |
| Massed Compute | 1× SXMH100 SXM5 (80GB) | not stated | $3.14 | $3.14 | not stated | 2026-11-12 | page |
| RunPod | 1× NVLSecure Cloud H100 NVL | per second | $3.19 | $3.19 | not stated | 2026-11-12 | page |
| Hyperstack | 1× SXMNVIDIA H100 SXM | per minute | $3.20 | $3.20 | not stated | 2026-11-12 | page |
| Thunder Compute | 1× SXMInstance H100 SXM | per minute | $3.20 | $3.20 | not stated | 2026-11-12 | page |
| Vast.aimarketplace | 1× SXMMarketplace median (H100 SXM) | not stated | $3.41 | $3.41 | not stated | 2026-11-12 | page |
| Runcrateours, live | 1× lowest live option | per second | $3.58 | $3.58 | 6 regions | 2026-11-12 | page |
| RunPod | 1× SXMSecure Cloud H100 SXM | per second | $3.99 | $3.99 | not stated | 2026-11-12 | page |
| DigitalOcean (Paperspace) | 1× HGXNVIDIA HGX H100 on-demand | per second | $4.41 | $4.41 | not stated | 2026-11-12 | page |
| Verda | 1× SXM1x H100 SXM5 80GB (30 vCPU, 120 GB RAM) | not stated | $5.09 | $5.09 | not stated | 2026-11-12 | page |
| Verda | 1× SXM1x H100 SXM5 80GB (32 vCPU, 170 GB RAM) | not stated | $5.09 | $5.09 | not stated | 2026-11-12 | page |
| Amazon Web Services | 1×p5.4xlarge | per second | $6.88 | $6.88 | us-east-1 | 2026-11-12 | page |
| Microsoft Azure | 1× NVLStandard_NC40ads_H100_v5 | not stated | $6.98 | $6.98 | eastus | 2026-11-12 | page |
| Oracle Cloud | 1×Oracle Cloud Infrastructure - Compute - GPU - H100 (B98415) | not stated | $10.00 | $10.00 | not stated | 2026-11-12 | page |
Announced for this size, not yet in effect
What the lowest, median and highest cost
derived: price x GPUs x hours| Per | Lowest | Median | Highest |
|---|---|---|---|
| GPU, per hour | $1.99 | $2.99 | $10.00 |
| GPU, 24 hours | $47.76 | $71.76 | $240.00 |
| GPU, 720-hour month | $1,432.80 | $2,152.80 | $7,200.00 |
The median column is the median of each provider's lowest price at this size ($2.99), not the price of any one provider. A month is 720 hours.
Per-GPU price across sizes
5 providers list 1× and another sizeOf 5 providers that list more than one size, 2 charge the same per GPU at every size, 0 charge less per GPU at their largest size, 3 charge more (comparing each provider's smallest listed size with its largest). The biggest move is Microsoft Azure, +76.1% per GPU.
| Provider | 1× $/GPU-hr | 2× $/GPU-hr | 4× $/GPU-hr | 8× $/GPU-hr | Smallest to largest |
|---|---|---|---|---|---|
| Massed Compute | $2.73 | $2.73 | $2.73 | $2.92 | +7.0% |
| Civo | $2.99 | none | none | $2.99 | same price |
| OVHcloud | $2.99 | $2.99 | $2.99 | none | +0.1% |
| Amazon Web Services | $6.88 | none | none | $6.88 | same price |
| Microsoft Azure | $6.98 | $6.98 | none | $12.29 | +76.1% |
- 2× lowest
- $2.73 per GPU at Massed Compute; the 1-GPU lowest of $1.99 is 27% below the 2-GPU lowest.
- 4× lowest
- $2.73 per GPU at Massed Compute; the 1-GPU lowest of $1.99 is 27% below the 4-GPU lowest.
- 8× lowest
- $2.30 per GPU at Vultr; the 1-GPU lowest of $1.99 is 13% below the 8-GPU lowest.
1-GPU H100: common questions
How much does a 1-GPU H100 configuration cost per hour?
$1.99 per hour at the lowest listing (Voltage Park, $1.99 per GPU). Among 13 providers the range is $1.99 to $10.00 per hour, with a median of $2.99.
Which variants are sold with 1 GPU?
variant not stated: $1.99 to $10.00 per GPU across 7 rows; PCIe: $2.50 to $2.89 per GPU across 2 rows; SXM: $2.95 to $5.09 per GPU across 8 rows; NVL: $3.11 to $6.98 per GPU across 3 rows; HGX: $4.41 to $4.41 per GPU across 1 row.
Which of these providers bill by the second?
Amazon Web Services, DigitalOcean (Paperspace), Runcrate, RunPod, and Thunder Compute state per-second billing on a 1-GPU row. Others either state per-minute or per-hour billing or none; the table shows each row's stated unit.
Are there marketplace or announced prices at this size?
8 marketplace rows (RunPod, Vast.ai, and TensorDock) are shown with a chip and left out of the statistics. 2 announced rows start later, the earliest on 2026-11-16.
Is a larger server lower per GPU than this?
The lowest 1-GPU price is $1.99. The lowest per-GPU prices at the other sizes are $2.73 at 2×, $2.73 at 4×, $2.30 at 8×.
Disclosure
H100 Rate Card is published by the Runcrate team, who also run Runcrate, an AI cloud. Runcrate is listed in the table for the H100, marked "ours, live". It is sorted by price like every other row, with no pinned position. Other providers' prices are read from their own public pricing pages. We have no paid placement.