GPU Comparison

NVIDIA H100 PCIevsNVIDIA B200

Detailed specifications, performance benchmarks, and pricing comparison to help you choose the right GPU for your AI workloads.

NVIDIA

NVIDIA H100 PCIe

Hopper · 2022

VRAM80 GB HBM3
FP3251 TFLOPS
FP16102 TFLOPS
Bandwidth2000 GB/s
TDP350W
Price$1.40/hr
NVIDIA

NVIDIA B200

Blackwell · 2024

Faster
VRAM192 GB HBM3e
FP3290 TFLOPS
FP16180 TFLOPS
Bandwidth8000 GB/s
TDP1000W
Price$3.40/hr

Specifications · Comparison

Side by side specs.

SpecNVIDIA H100 PCIeNVIDIA B200
ArchitectureHopperBlackwell
VRAM80 GB192 GB+58.3%
Memory TypeHBM3HBM3e
Memory Bandwidth2000 GB/s8000 GB/s+75%
FP32 Performance51 TFLOPS90 TFLOPS+43.3%
FP16 Performance102 TFLOPS180 TFLOPS+43.3%
INT8 Performance2040 TOPS3600 TOPS+43.3%
TDP350W1000W
Form FactorPCIeSXM
Price (avg/hr)$1.40$3.40

Performance · Analysis

Performance breakdown.

Compute (FP32)

Raw single-precision floating point throughput

NVIDIA H100 PCIe
51TFLOPS
NVIDIA B200
90TFLOPSWinner

NVIDIA B200 is 43.3% faster

Training (FP16)

Half-precision performance for deep learning training

NVIDIA H100 PCIe
102TFLOPS
NVIDIA B200
180TFLOPSWinner

NVIDIA B200 is 43.3% faster

Inference (INT8)

Integer performance for model inference workloads

NVIDIA H100 PCIe
2040TOPS
NVIDIA B200
3600TOPSWinner

NVIDIA B200 is 43.3% faster

Memory Bandwidth

Data transfer rate between memory and compute units

NVIDIA H100 PCIe
2000GB/s
NVIDIA B200
8000GB/sWinner

NVIDIA B200 is 75% faster

Best Compute

NVIDIA B200

Most Memory

NVIDIA B200

Best Training

NVIDIA B200

Best Value

NVIDIA H100 PCIe

Pricing · Cost

Cost comparison.

Hourly

NVIDIA H100 PCIe
$1.40/hr
NVIDIA B200
$3.40/hr

Save $2.00 with NVIDIA H100 PCIe

Daily

NVIDIA H100 PCIe
$33.60/day
NVIDIA B200
$81.60/day

Save $48.00 with NVIDIA H100 PCIe

Monthly

NVIDIA H100 PCIe
$1008.00/mo
NVIDIA B200
$2448.00/mo

Save $1440.00 with NVIDIA H100 PCIe

Use Cases · Workloads

Best for your workload.

NVIDIA

NVIDIA H100 PCIe

Deep Learning Training
AI Inference
Data Analytics
NVIDIA

NVIDIA B200

Large Language Models
Multi-Modal AI
Scientific Computing
Enterprise AI

Platform · Benefits

Why Runcrate.

Instant Deployment

Get your GPU instance running in minutes with pre-configured AI environments. No setup complexity.

Pay Per Hour

Only pay for the compute you actually use. Prepaid credits with transparent, per-hour billing.

Reliable Infrastructure

Enterprise-grade reliability with automatic failover and data persistence across sessions.

FAQ · Questions

Common questions.

Start Building Today

Deploy NVIDIA H100 PCIe or NVIDIA B200

Get started with GPU cloud computing in minutes. No setup complexity, no long-term commitments.

No credit card required
Pay-per-hour billing
Cancel anytime