NVIDIA Hopper

Industry standard (2022-2024)

Hopper GPU Family

NVIDIA Hopper is the industry standard for AI training, featuring HBM3 memory, the Transformer Engine for LLM workloads, and NVLink 4 interconnects (900 GB/s). The H100 is the most widely-deployed AI GPU in cloud computing today.

6 GPU models in this family

GPU Comparison

GPU ModelVRAMBandwidthInterconnectBest ForStarting PriceTDPYear
H200 PCIe
PCIe Gen5
141 GB HBM3e4.8 TB/sPCIe Gen5 (128 GB/s) — no NVLinkLLM inference350W2024View
H200 SXM
SXM5
141 GB HBM3e4.8 TB/sNVLink 4 (900 GB/s)LLM training (70B+ params)700W2024View
GH200
Superchip (Grace CPU + Hopper GPU)
144 GB HBM3e (or 96 GB HBM3)4.8 TB/s (HBM3e)NVLink 4 (900 GB/s)Large-scale AI training1,000W2023View
H100 NVL
PCIe Gen5 (dual-slot)
94 GB HBM33.9 TB/sNVLink 4 (900 GB/s) — PCIe card with NVLink bridgeAI training350W2023View
H100 PCIe
PCIe Gen5
80 GB HBM33.35 TB/sPCIe Gen5 (128 GB/s) — no NVLinkAI inference350W2022View
H100 SXM
SXM5
80 GB HBM33.35 TB/sNVLink 4 (900 GB/s)AI training700W2022View