H200 SXM
LLM inference · fine-tuning
$1,832/ 30 days
View GPU H200 SXM vs A100 SXM
Compare memory, node configurations, and fixed weekly and monthly cost before choosing between NVIDIA H200 SXM and NVIDIA A100 SXM.
H200 SXM provides 61 GB more VRAM per card. A100 SXM costs $1,040 less per 30-day term . Choose based on whether memory headroom or lower fixed cost is the binding constraint.
LLM inference · fine-tuning
training · batch inference
| Specification | H200 SXM | A100 SXM |
|---|---|---|
| Architecture | Hopper | Ampere |
| VRAM | 141 GB HBM3e | 80 GB HBM2e |
| Host resources / card | 24 vCPU · 192 GB RAM | 20 vCPU · 160 GB RAM |
| Local NVMe / card | 3.84 TB | 2 TB |
| 7-day price / card | $512 | $224 |
| 30-day price / card | $1,832 | $792 |
| 4-GPU node / 30 days | $6,742 | $2,915 |
| Catalog quantity | 265 cards | 780 cards |