NVIDIA H200 SXM
Data center GPU · Hopper
Summary
The most powerful variant of NVIDIA's H200 GPU – ideal for multi-GPU training runs.
Launched
Q3 2024
VRAM
141 GB
Mem. bandwidth
4,800 GB/s
On-demand from
Tech specs
VRAM 141 GB HBM3e
Memory bandwidth 4,800 GB/s
Interface SXM
CUDA cores 16,896
Tensor cores 528 (Gen 4)
TDP 700 W
Supported data types
FP64FP32FP16BF16FP8INT8
Cloud rental prices
15 providers · available in 26 countries · lowest price per GPU, per hour
On-demand
$2.97 – $10.60 /GPU/h
Reserved
$2.53 – $7.31 /GPU/h
Spot
$1.40 – $5.36 /GPU/h
| Provider | On-demand | Reserved | Spot |
|---|---|---|---|
 AceCloud | — | $3.61 | — |
 AWS | $7.91 | — | — |
| | $10.60 | $4.66 | $2.12 |
 CoreWeave | $6.31 | — | $2.58 |
 Crusoe | $4.29 | — | — |
| | $10.60 | $4.65 | $5.36 |
 Hyperstack | $3.99 | $2.79 | — |
 Nebius | $4.50 | — | $2.45 |
 Oracle | $10.00 | — | — |
 Runpod | $4.39 | — | — |
 Seeweb | $2.97 | $2.53 | — |
| | $3.78 | — | — |
 Together AI | $5.99 | — | — |
 Vast.ai | $4.08 | — | $3.48 |
 Verda | $4.00 | $3.68 | $1.40 |
Models that fit in VRAM
Open-weights models that fit in VRAM. Estimated using 🤗 accelerate, plus approximation for up to 8K context.
Media
Introducing Mistral 3 (Mistral Large 3 trained on 3,000 NVIDIA H200 GPUs) CoreWeave First to Market with NVIDIA H200 Tensor Core GPUs, Ushering in a New Era of AI Infrastructure Performance Hippocratic AI — NVIDIA Customer Stories (patient-facing healthcare agents on H200) xAI Colossus adds 50,000 H200 GPUs Nebius launches among the first H200 clusters in Europe and the US
Peak theoretical performance
FP8 Tensor Core 1,979 TFLOPS
INT8 Tensor Core 1,979 TOPS
BF16 Tensor Core 990 TFLOPS
FP16 Tensor Core 990 TFLOPS
TF32 Tensor Core 494 TFLOPS
FP32 67 TFLOPS
FP64 34 TFLOPS
FP64 Tensor Core 67 TFLOPS
Performance figures assume no sparsity; in cases where only sparse performance figures are published by the manufacturer, these are halved to give approximate dense performance.
Resources
Detailed documentation from the manufacturer.
Similar GPUs
Other accelerators you might compare.