AVAILABLE NOW · CAD PRICING

NVIDIA
H200 SXM
GPU CLOUD

The world's most deployed AI GPU. 141GB HBM3e, 989 TFLOPS FP16, 4.8 TB/s memory bandwidth — on AWS and Azure, billed per hour in Canadian dollars.

Deploy H100 Now → Get Enterprise Quote
Starting From
$4.10
CAD / HR / GPU
≈ $2.29 USD at current rates
No contracts or minimums
Hourly billing in CAD
Deploy in under 60 seconds
99.9% uptime SLA
AWS & Azure infrastructure
H200 SXM · $4.10 H200 SXM · $4.10 B300 · $6.71 GB300 NVL72 · $409
// Technical Specifications

H200 SXM
FULL SPECS

SPECIFICATIONH200 SXM
ArchitectureNVIDIA Hopper
GPU Memory141 GB HBM3e
Memory Bandwidth4.8 TB/s
FP16 / BF16 Performance989 TFLOPS
FP8 Performance1,979 TFLOPS
TF32 Performance494 TFLOPS
FP64 Performance33.5 TFLOPS
NVLink Bandwidth900 GB/s
PCIe Bandwidth128 GB/s
TDP (Power)700W
Transformer EngineYes — 2nd Gen
CUDA Cores16,896
SScoreCompute Price$4.10 CAD/hr
Cloud PlatformAWS + Azure
// Ideal Workloads

WHAT THE H100
IS BUILT FOR

🤖
LLM INFERENCE
Serve LLaMA, Mistral, GPT-class models at scale with vLLM. H100's Transformer Engine and HBM3 memory deliver industry-leading token throughput.
🎯
MODEL FINE-TUNING
Fine-tune 7B–70B models with LoRA or full-parameter training. 80GB VRAM fits most open-source models without quantization.
🔍
RAG PIPELINES
Power high-throughput retrieval-augmented generation. Embed, retrieve, and generate at production scale with sub-200ms latency.
🤝
MULTI-AGENT SYSTEMS
Run parallel agent inference loops — planning, tool use, and reasoning — across multiple H100s with NVLink for fast inter-GPU communication.
🖼️
IMAGE & VIDEO AI
Stable Diffusion, FLUX, video generation, and multimodal models run efficiently on 80GB HBM3 with high-bandwidth memory access.
📊
ML TRAINING
Train models up to ~13B parameters on a single H100. Scale to multi-GPU with NVLink. Ideal for research and production training runs.
// Performance

H100 BENCHMARK
RESULTS

Real-world throughput on SScoreCompute H200 SXM instances. LLaMA-3 70B, batch size 8.

LLM Inference — Output Tokens / Second
LLaMA-3 70B (FP16)
FP16
1,680tok/s
LLaMA-3 70B (FP8)
FP8
2,240tok/s
Mistral 7B (FP16)
FP16
8,400tok/s
Training Throughput
GPT-3 style 6.7B (BF16)
BF16
14,200tok/s
Time to First Token
TTFT
42ms

READY TO DEPLOY
H100 GPUS?

On-demand, by the hour, in CAD. No contracts. No waitlists. Live in 60 seconds.

Deploy H100 Now → Get Enterprise Pricing