NVIDIA H200 cloud GPU prices
Compare listed H200 rental prices for memory-intensive AI workloads. Check the minimum GPU count alongside the headline per-GPU rate before planning a budget.
Rates dated Nov 10, 2026–Nov 10, 2026
Publication dates, not live quotes. Confirm the current rate and capacity with the provider.
Catalog generated .
Lowest listed on-demand rate
$5.29/GPU·hr
RunPod · H200 · Unspecified · 141 GB per GPU · 1 GPUs · $5.29/instance·hr · AP-JP-1
Lowest listed spot / interruptible rate
$0.847/GPU·hr
AWS · H200 · Unspecified · 141 GB per GPU · 8 GPUs · $6.77/instance·hr · eu-north-1
Compare rental prices
Lowest listed compute rates. Explore comparison filters for your region.
| Provider / GPU / variant | VRAM per GPU | Pricing | Per GPU / hr | Full instance / hr | Region | Provider / rate date |
|---|---|---|---|---|---|---|
| AWS H200 · Unspecified Instance: p5en.48xlarge Link to offer | 141 GB | Spot / interruptible | $0.847 | $6.77 8 GPUs | eu-north-1 Stock unknown Stock & project cost | Check provider priceRate dated |
| Nebius H200 · SXM Instance: gpu-h200-sxm 1gpu-16vcpu-200gb Link to offer | 141 GB | Spot / interruptible | $1.20 | $1.20 1 GPUs | eu-north1 Stock unknown Stock & project cost | Check provider priceRate dated |
| Nebius H200 · SXM Instance: gpu-h200-sxm 8gpu-128vcpu-1600gb Link to offer | 141 GB | Spot / interruptible | $1.20 | $9.60 8 GPUs | eu-north1 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 1H200.141S.44V Link to offer | 141 GB | Spot / interruptible | $3.22 | $3.22 1 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 2H200.141S.88V Link to offer | 141 GB | Spot / interruptible | $3.22 | $6.44 2 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 4H200.141S.176V Link to offer | 141 GB | Spot / interruptible | $3.22 | $12.88 4 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 8H200.141S.176V Link to offer | 141 GB | Spot / interruptible | $3.22 | $25.76 8 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| RunPod H200 · Unspecified Instance: NVIDIA H200 Link to offer | 141 GB | On-demand | $5.29 | $5.29 1 GPUs | AP-JP-1 Stock unknown Stock & project cost | Check provider priceRate dated |
| RunPod H200 · Unspecified Instance: NVIDIA H200 Link to offer | 141 GB | On-demand | $5.29 | $10.58 2 GPUs | EUR-IS-5 Stock unknown Stock & project cost | Check provider priceRate dated |
| Nebius H200 · SXM Instance: gpu-h200-sxm 1gpu-16vcpu-200gb Link to offer | 141 GB | On-demand | $5.40 | $5.40 1 GPUs | eu-north1 Stock unknown Stock & project cost | Check provider priceRate dated |
| Nebius H200 · SXM Instance: gpu-h200-sxm 8gpu-128vcpu-1600gb Link to offer | 141 GB | On-demand | $5.40 | $43.20 8 GPUs | eu-north1 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 1H200.141S.44V Link to offer | 141 GB | On-demand | $6.44 | $6.44 1 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 2H200.141S.88V Link to offer | 141 GB | On-demand | $6.44 | $12.88 2 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 4H200.141S.176V Link to offer | 141 GB | On-demand | $6.44 | $25.76 4 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| DataCrunch H200 · Unspecified Instance: 8H200.141S.176V Link to offer | 141 GB | On-demand | $6.44 | $51.52 8 GPUs | FIN-01 Stock unknown Stock & project cost | Check provider priceRate dated |
| AWS H200 · Unspecified Instance: p5en.48xlarge Link to offer | 141 GB | On-demand | $7.91 | $63.30 8 GPUs | us-east-1 Stock unknown Stock & project cost | Check provider priceRate dated |
| Oracle Cloud H200 · Unspecified Instance: BM.GPU.H200.8 Link to offer | 141 GB | On-demand | $10.00 | $80.00 8 GPUs | af-casablanca-1 Stock unknown Stock & project cost | Check provider priceRate dated |
| Azure H200 · Unspecified Instance: Standard_ND96isr_H200_v5 Link to offer | 141 GB | On-demand | $10.60 | $84.80 8 GPUs | eastus2 Stock unknown Stock & project cost | Check provider priceRate dated |
| Azure H200 · Unspecified Instance: Standard_ND96isr_H200_v5 Link to offer | 141 GB | Spot / interruptible | $10.60 | $84.80 8 GPUs | eastus2 Stock unknown Stock & project cost | Check provider priceRate dated |
Manufacturer specifications, with conditions
Reviewed September 26, 2026 against NVIDIA documentation. These are hardware specifications and theoretical peaks, not NVGPU measurements or third-party benchmark results. Sparse Tensor Core peaks require supported structured sparsity; they are not expected dense training throughput.
| Reference variant | Memory | Memory bandwidth | FP16 Tensor Core peak | Conditions and source |
|---|---|---|---|---|
| H200 SXM | 141 GB HBM3e | 4.8 TB/s | Not compared here | HGX reference configuration. Confirm the rental SKU and reported memory; memory capacity alone is not a throughput result. NVIDIA source |
NVGPU has not run application benchmarks for these comparisons. To assess a third-party result, require the model and revision, precision, batch size, sequence lengths, GPU count and variant, software versions, interconnect, test date and methodology.
Before you rent
- Compare the reported per-GPU VRAM against your model and KV-cache requirements.
- Confirm whether the advertised configuration can be rented as a single GPU or only as a larger instance.
- Benchmark your actual inference or training stack before committing to a longer reservation.
What is included?
These are listed compute rates. The instance shown is the configuration being priced, not evidence of the provider's smallest available rental. Storage, data transfer, tax, licenses and support can be additional. Reserved contracts, commitment discounts and serverless per-request charges are not compared. Listed spot capacity can disappear or be interrupted.
Hourly price vs. total cost
Multiply the full-instance hourly rate by billed hours, then add storage, transfer and other charges. Compare the same workload and quality target. Rental throughput and uptime have not been measured by NVGPU.