Skip to main content

H100 vs T4

The H100 delivers 15.2× the FP16 compute of the T4 while costing 9.9× more on average, making it 54% more cost-efficient per TFLOP.

Spec Comparison

Spec
H100
T4
Architecture
Hopper
Older
VRAM
80 GB
16 GB
FP16 TFLOPS
989
65
Memory Bandwidth
3,350 GB/s
320 GB/s
MSRP
$30,000
N/A
Used/Market Price
N/A
$1,500

Live Pricing

H100
Cheapest
$0.801/hr
Median
$8.97/hr
Average
$16/hr
Providers
36
Instances
327
T4
Cheapest
$0.0684/hr
Median
$0.909/hr
Average
$1.53/hr
Providers
8
Instances
1557

Cost Efficiency

Metric
H100
T4
$/TFLOP/hr
$0.0091
$0.0140
$/GB-VRAM/hr
$0.1121
$0.0568

Lower $/TFLOP and $/GB-VRAM is better. Based on median hourly pricing.

Provider Availability

H100 Cloud Pricing

327 instances
ProviderGPUsVRAMRegionTypePrice/hr
CHEAPESTVerda1x80 GBEU-CentralSpot$0.801Rent
Verda1x80 GBEU-CentralSpot$0.801Rent
Nebius1x80 GBeu-north1Spot$1.25Rent
GCP1x80 GBeurope-north1-cSpot$1.29Rent
GCP1x80 GBEU-WestSpot$1.39Rent
Vast.ai1x80 GBfr-franceSpot$1.47Rent
Vast.ai1x80 GBfr-franceOn-Demand$1.55Rent
Verda2x160 GBEU-CentralSpot$1.60Rent
Shadeform1x80 GBUS-EastOn-Demand$1.66Rent
E2E Networks1x80 GBAsia-PacificOn-Demand$1.69Rent

T4 Cloud Pricing

1557 instances
ProviderGPUsVRAMRegionTypePrice/hr
CHEAPESTAzure1x16 GBAsia-PacificSpot$0.0684Rent
Azure1x16 GBUS-CentralSpot$0.0686Rent
Azure1x16 GBAsia-PacificSpot$0.0977Rent
Azure1x16 GBUS-CentralSpot$0.0980Rent
GCP1x16 GBme-west1-bSpot$0.112Rent
GCP1x16 GBme-west1-cSpot$0.112Rent
GCP1x16 GBEU-WestSpot$0.117Rent
GCP1x16 GBme-west1-bSpot$0.118Rent
GCP1x16 GBme-west1-cSpot$0.118Rent
AWS1x16 GBSouth AmericaSpot$0.123Rent

Compare with Other GPUs