H200 vs GH200
H200 starts at $2.60/GPU-hour on-demand; GH200 starts at $2.29/GPU-hour on-demand. Compare the exact variants, memory and published specifications below. Peak compute figures do not establish application speed.
H200
GH200
Spec sheet
| H200 | GH200 | |
|---|---|---|
| Spec variant | H200 (SXM) | GH200 |
| Architecture | Hopper | Hopper |
| VRAM (spec variant) | 141 GB | 96 GB |
| FP16 TFLOPS (dense) | 989 | 989 |
| Memory bandwidth | 4800 GB/s | 4000 GB/s |
| TDP | 700 W | 700 W |
| Interconnect | NVLink 900GB/s | NVLink-C2C 900GB/s |
| Launched | 2023 | 2023 |
Rental price per manufacturer-rated FP16 compute: H200 $2.63 per PFLOP-hour vs GH200 $2.32. Formula: cheapest $/hr ÷ dense FP16 PFLOPS (1 PFLOP = 1000 TFLOPS). This uses H200 (SXM) and GH200, whose datasheets may differ from the variants above. It is not measured job throughput. Specs: H200 manufacturer, GH200 manufacturer.
What they cost to rent (2026-10-02)
| H200 | GH200 | |
|---|---|---|
| Cheapest on-demand $/hr | $2.60 | $2.29 |
| Cheapest spot/community $/hr | $2.99 | – |
| Cheapest source | GMI Cloud | Lambda |
| Providers offering it | 22 | 3 |
Lowest recorded on-demand rate per configuration; highlighted = cheaper. Spot/marketplace tiers shown on the individual pages.
FAQ
GH200 is cheaper: the H200 rents from $2.60 per GPU-hour and the GH200 from $2.29 on-demand, updated 2026-10-02.
The largest recorded H200 variant has 143 GB and the largest GH200 variant has 96 GB. Other variants may have less memory. The H200 has more.
Compare more rental options: the screener · price movers · methodology.