…
Cloud rental price, compute per dollar and specs, side by side.
Open in the interactive comparer →| Spec / price | H100 SXM | H200 SXM |
|---|---|---|
| Memory | 80 GB HBM3 | 141 GB HBM3ebetter |
| Memory bandwidth | 3,350 GB/s | 4,800 GB/sbetter |
| FP16 / BF16 (dense) | 0.989 PF | 0.989 PF |
| FP8 (dense) | 1.98 PF | 1.98 PF |
| FP4 (dense) | — | — |
| INT8 (dense) | 1.98 PF | 1.98 PF |
| TF32 (dense) | 0.495 PF | 0.495 PF |
| FP32 (dense) | 0.067 PF | 0.067 PF |
| FP64 (dense) | 0.067 PF | 0.067 PF |
| Power | 700 W | 700 W |
| Architecture | Hopper (2022) | Hopper (2024) |
| Lowest on-demand | $1.66better | $2.10 |
| Median on-demand | $3.19better | $3.99 |
| Lowest spot | $0.79 | $0.79 |
| FP16/BF16 $ per PFLOP·h | $1.68better | $2.12 |
| Providers | 28 | 22 |
The H100 SXM and H200 SXM have the same dense FP16/BF16 compute. The H200 SXM has 1.4× more memory bandwidth than the H100 SXM. Real-world speedups depend on the workload: training and prefill scale with compute, while LLM decoding scales mostly with memory bandwidth.
Right now the H100 SXM is cheaper per unit of compute: $1.68 vs $2.12 per PFLOP-hour at the lowest on-demand prices.
Running one GPU around the clock at the lowest on-demand price costs about $1,212 per month for the H100 SXM and $1,533 for the H200 SXM.
The H200 SXM has more memory: 141 GB vs 80 GB.