Kovara
Loading your page…
You can continue using the navigation while this page loads.
You can continue using the navigation while this page loads.
A100 80GB SXM has 17% less memory than GH200 96GB HBM3 (80 GB vs 96 GB). Its memory bandwidth is 49% less (2.039 vs 4 TB/s). In the cloud, the median on-demand A100 80GB SXM price is $1.69 per GPU-hour against $5.19 for the GH200 96GB HBM3. Per gigabyte of GPU memory, that is $0.021 vs $0.054 per hour.
Manufacturer figures as recorded by Kovara. Higher memory and bandwidth are highlighted; neither is a benchmark of your workload.
| A100 80GB SXM | GH200 96GB HBM3 | |
|---|---|---|
| Manufacturer | NVIDIA | NVIDIA |
| Architecture | Ampere | Hopper + Grace |
| Memory | 80 GB HBM2E | 96 GB HBM3 |
| Memory bandwidth | 2.039 TB/s | 4 TB/s |
| Power (TDP) | 400 W | 900 W |
| Form factor | SXM | Grace Hopper Superchip |
Public offers Kovara tracks, normalised to US dollars per GPU-hour. Prices use on-demand terms where available; spot and reserved offers are on each GPU’s price pages. Lower is highlighted.
| A100 80GB SXM | GH200 96GB HBM3 | |
|---|---|---|
| Lowest price | $1.38 | $2.29 |
| Median price | $1.69 | $5.19 |
| Highest price | $3.18 | $6.50 |
| Median per GB of memory | $0.021 | $0.054 |
| Public offers (all terms) | 20 | 5 |
| Providers | 9 | 3 |
Lowest tracked price per provider. Open a provider to see every offer, region and term.
Right now the cheapest on-demand A100 80GB SXM Kovara tracks is $1.38 per GPU-hour and the cheapest GH200 96GB HBM3 is $2.29. Medians are $1.69 and $5.19. The cheaper GPU per hour is not always cheaper per job: a faster or larger GPU can finish sooner or need fewer units.
The GH200 96GB HBM3 has more memory: 96 GB against 80 GB. Memory decides which models and batch sizes fit on one GPU.
No. This page compares manufacturer specifications and public prices. Real performance depends on the model, precision, software and interconnect. Kova can help you estimate cost for a specific workload.