NVIDIA H200 SXM

Cloud GPU Pricing — Live from 8+ Providers
Starting From
$2.090/hr
8 on-demand listings
VRAM
141 GB HBM3e
Architecture
Hopper
FP16 Performance
1,979 TFLOPS
Memory Bandwidth
4.8 TB/s
Listings
11 options
Provider Pricing
ProviderPrice / hrEst. MonthlyTypeVRAMAction
BestSesterce $2.0900/hr $1,525.7/mo On-Demand 141 GB Rent Now →
FluidStack $3.2900/hr $2,401.7/mo On-Demand 141 GB Rent Now →
Hyperstack $3.5000/hr $2,555/mo On-Demand 141 GB Rent Now →
Vast.ai $3.5565/hr $2,596.25/mo On-Demand 140 GB Rent Now →
RunPod $3.7900/hr $2,766.7/mo On-Demand 143 GB Rent Now →
Lambda Labs $3.9900/hr $2,912.7/mo On-Demand 141 GB Rent Now →
RunPod $4.3900/hr $3,204.7/mo On-Demand 141 GB Rent Now →
CoreWeave $5.4900/hr $4,007.7/mo On-Demand 141 GB Rent Now →
RunPod $0.5000/hr $365/mo Spot 143 GB Rent Now →
Vast.ai $1.0414/hr $760.22/mo Spot 140 GB Rent Now →
RunPod $3.5900/hr $2,620.7/mo Spot 141 GB Rent Now →
About the NVIDIA H200 SXM

The NVIDIA H200 SXM is the most powerful GPU available, featuring 141 GB of HBM3e with 4.8 TB/s bandwidth — the largest GPU memory pool on the market. Designed for frontier models and mixture-of-experts architectures that exceed what any other GPU can hold. The choice for next-generation AI research.

GPU Specifications
VRAM141 GB HBM3e
ArchitectureHopper
FP16 Performance1,979 TFLOPS
Memory Bandwidth4.8 TB/s
TDP700W
InterconnectNVLink 4.0
Best For
Frontier LLM TrainingMixture of Experts ModelsMulti-modal ModelsHPC Research

Compare All NVIDIA H200 SXM Options Side by Side

Filter by price type, sort by cost, and find the best deal for your workload.

Open Full Comparison →