Skip to content
GPUVerse Blog

NVIDIA · Hopper · H200

H200

141 GB of HBM3e at 4.8 TB/s, built for training, inference, rag.

From $2.89/hrindicative · across 10 providers

H200 render

Specifications

Memory

141 GB

HBM3e

Bandwidth

4.80 TB/s

FP16 tensor

1979 TFLOPS

FP8 tensor

3958 TFLOPS

Interconnect

NVLink

TDP

700 W

CUDA cores

16,896

Tensor cores

528

Where to rent it

ProviderTypeSpotDetails
RunPod logoRunPodneocloudYesView
Vast.ai logoVast.aimarketplaceYesView
CoreWeave logoCoreWeaveneocloud-View
Lambda logoLambdaneocloud-View
Nebius logoNebiusneocloud-View
Crusoe logoCrusoeneocloud-View
Amazon AWS logoAmazon AWShyperscalerYesView
Google Cloud logoGoogle CloudhyperscalerYesView
Microsoft Azure logoMicrosoft AzurehyperscalerYesView
Oracle Cloud logoOracle CloudhyperscalerYesView

Compare with

H100 SXM render
NVIDIA · Hopper

H100 SXM

Memory
80 GB HBM3
Bandwidth
3.4 TB/s
highfrom$1.99/hr
B200 render
NVIDIA · Blackwell

B200

Memory
192 GB HBM3e
Bandwidth
8.0 TB/s
flagshipfrom$5.49/hr
GB200 render
NVIDIA · Blackwell

GB200

Memory
192 GB HBM3e
Bandwidth
8.0 TB/s
flagshipfrom$6.99/hr
H100 PCIe render
NVIDIA · Hopper

H100 PCIe

Memory
80 GB HBM3
Bandwidth
2.0 TB/s
highfrom$2.49/hr