NVIDIA · Hopper Architecture

Umuupa ng NVIDIA H200 SXM sa Cloud

Enhanced H100 with 141GB HBM3e memory and 43% more bandwidth. Ideal for models that need more VRAM than the H100 provides.

VRAM 141 GB HBM3e
Bandwidth 4,800 GB/s
FP16 990.0 TFLOPS
FP32 67.0 TFLOPS
TDP 700W
Arkitektura Hopper
Pinakamurang On-Demand $2.05/hr
Karaniwang On-Demand $3.03/hr
Mga Provider 3

Ihambing ang Presyo ng NVIDIA H200 SXM Cloud — 3 Providers

On-Demand

NVIDIA H200 SXM — On-Demand pricing
Provider Presyo / GPU / oras Availability Mga Tala Aksyon
Vast.ai United States $2.05/hr PINAKAMURA Available Marketplace avg Bisitahin ang Provider
DigitalOcean United States $3.44/hr Available HGX H200 Bisitahin ang Provider
RunPod United States $3.59/hr Available Secure Cloud Bisitahin ang Provider

Spot / Preemptible

NVIDIA H200 SXM — Spot / Preemptible pricing
Provider Presyo / GPU / oras Availability Mga Tala Aksyon
RunPod United States $3.59/hr PINAKAMURA Available Community Cloud Bisitahin ang Provider

Huling beripikasyon ng presyo: April 13, 2026

Teknikal na Espesipikasyon ng NVIDIA H200 SXM

Tagagawa NVIDIA
Arkitektura Hopper
VRAM 141 GB HBM3e
Bandwidth 4,800 GB/s
FP16 (Tensor) 990.0 TFLOPS
FP32 67.0 TFLOPS
TDP 700W
Taon ng Paglabas 2024
Segmento Data center
Uri ng Memorya HBM3e

Pinakamainam Para sa

Large model training inference for 70B+ parameter models

Ihambing sa Ibang GPU

Tingnan kung paano tumitimbang ang NVIDIA H200 SXM laban sa iba pang popular na cloud GPUs sa specs, presyo, at availability.