NVIDIA H100 SXM

Cloud GPU Pricing — Live from 8+ Providers
Starting From
$1.830/hr
10 on-demand listings
VRAM
80 GB HBM3
Architecture
Hopper
FP16 Performance
1,979 TFLOPS
Memory Bandwidth
3.35 TB/s
Listings
12 options
Provider Pricing
ProviderPrice / hrEst. MonthlyTypeVRAMAction
BestSesterce $1.8300/hr $1,335.9/mo On-Demand 80 GB Rent Now →
Verda $1.9900/hr $1,452.7/mo On-Demand 80 GB Rent Now →
FluidStack $2.1000/hr $1,533/mo On-Demand 80 GB Rent Now →
Salad Cloud $2.2000/hr $1,606/mo On-Demand 80 GB Rent Now →
Vast.ai $2.3622/hr $1,724.41/mo On-Demand 80 GB Rent Now →
Hyperstack $2.4000/hr $1,752/mo On-Demand 80 GB Rent Now →
Lambda Labs $2.4900/hr $1,817.7/mo On-Demand 80 GB Rent Now →
Vultr $2.5400/hr $1,854.2/mo On-Demand 80 GB Rent Now →
RunPod $3.2900/hr $2,401.7/mo On-Demand 80 GB Rent Now →
CoreWeave $4.7600/hr $3,474.8/mo On-Demand 80 GB Rent Now →
Vast.ai $1.4422/hr $1,052.81/mo Spot 80 GB Rent Now →
RunPod $2.6900/hr $1,963.7/mo Spot 80 GB Rent Now →
About the NVIDIA H100 SXM

The NVIDIA H100 SXM is the gold standard for AI training. With 80 GB of HBM3 memory and 3.35 TB/s bandwidth, it is the preferred GPU for training large language models (LLMs). The SXM form factor enables NVLink multi-GPU configurations for scaling to hundreds of GPUs — the workhorse behind GPT-4 class model training.

GPU Specifications
VRAM80 GB HBM3
ArchitectureHopper
FP16 Performance1,979 TFLOPS
Memory Bandwidth3.35 TB/s
TDP700W
InterconnectNVLink 4.0
Best For
LLM TrainingLarge Model InferenceScientific HPCDeep Learning Research

Compare All NVIDIA H100 SXM Options Side by Side

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

Open Full Comparison →