H200 vs RTX5090
The H200 delivers 9.4× the FP16 compute of the RTX5090 while costing 7.5× more on average, making it 25% more cost-efficient per TFLOP.
Spec Comparison
Spec
H200
RTX5090
Architecture
Hopper
Blackwell
VRAM
141 GB
32 GB
FP16 TFLOPS
1,979
210
Memory Bandwidth
4,800 GB/s
1,792 GB/s
MSRP
$40,000
$1,999
Used/Market Price
N/A
N/A
Live Pricing
Cost Efficiency
Metric
H200
RTX5090
$/TFLOP/hr
$0.0060
$0.0076
$/GB-VRAM/hr
$0.0849
$0.0497
Lower $/TFLOP and $/GB-VRAM is better. Based on median hourly pricing.
Provider Availability
H200 Cloud Pricing
133 instances| Provider | GPUs | VRAM | Region | Type | Price/hr | |
|---|---|---|---|---|---|---|
| CHEAPESTVerda | 1x | 141 GB | EU-Central | Spot | $1.19 | Rent |
| Nebius | 1x | 141 GB | eu-north1 | Spot | $1.45 | Rent |
| Nebius | 1x | 141 GB | EU-West | Spot | $1.45 | Rent |
| Nebius | 1x | 141 GB | US-Central | Spot | $1.45 | Rent |
| RunPod | 1x | 141 GB | US-East | Spot | $2.29 | Rent |
| RunPod | 1x | 141 GB | EU-FR-1 | Spot | $2.29 | Rent |
| RunPod | 1x | 141 GB | EUR-IS-2 | Spot | $2.29 | Rent |
| RunPod | 1x | 141 GB | EUR-IS-4 | Spot | $2.29 | Rent |
| RunPod | 1x | 141 GB | US-CA-2 | Spot | $2.29 | Rent |
| RunPod | 1x | 141 GB | US-GA-2 | Spot | $2.29 | Rent |
RTX5090 Cloud Pricing
108 instances| Provider | GPUs | VRAM | Region | Type | Price/hr | |
|---|---|---|---|---|---|---|
| CHEAPESTVast.ai | 1x | 32 GB | US-East | Spot | $0.0940 | Rent |
| Vast.ai | 1x | 32 GB | Asia-Pacific | Spot | $0.100 | Rent |
| Vast.ai | 1x | 32 GB | EU-Central | Spot | $0.120 | Rent |
| Vast.ai | 2x | 64 GB | Asia-Pacific | Spot | $0.267 | Rent |
| Vast.ai | 1x | 32 GB | US-Central | Spot | $0.268 | Rent |
| Vast.ai | 1x | 32 GB | US-Central | On-Demand | $0.280 | Rent |
| Vast.ai | 1x | 32 GB | US-West | Spot | $0.293 | Rent |
| Vast.ai | 1x | 32 GB | EU-Central | On-Demand | $0.330 | Rent |
| Vast.ai | 1x | 32 GB | Asia-Pacific | On-Demand | $0.368 | Rent |
| Vast.ai | 2x | 64 GB | vn-vietnam | Spot | $0.373 | Rent |