GH200 vs H100
Datasheet specs side by side, plus what each card actually costs to rent today across every provider we track.
GH200
H100
Spec sheet
| GH200 | H100 | |
|---|---|---|
| Architecture | Hopper | Hopper |
| VRAM (max variant) | 96 GB | 94 GB |
| FP16 TFLOPS (dense) | 989 | 989 |
| Memory bandwidth | 4000 GB/s | 3350 GB/s |
| TDP | 700 W | 700 W |
| Interconnect | NVLink-C2C 900GB/s | NVLink 900GB/s |
| Launched | 2023 | 2022 |
Computed price-performance at today's cheapest on-demand rates: GH200 $2.32 per PFLOP-hour vs H100 $2.01. Formula: cheapest $/hr ÷ dense FP16 PFLOPS (1 PFLOP = 1000 TFLOPS). Specs from vendor datasheets (source, source).
What they cost to rent (2026-08-22)
| GH200 | H100 | |
|---|---|---|
| Cheapest on-demand $/hr | $2.29 | $1.99 |
| Cheapest spot/community $/hr | – | $0.94 |
| Cheapest source | Lambda | Voltage Park |
| Providers offering it | 2 | 32 |
Cheapest verified on-demand rate per side from today's snapshot; highlighted = cheaper. Spot/marketplace tiers shown on the individual pages.
FAQ
Is the GH200 cheaper to rent than the H100?
H100 is cheaper: the GH200 rents from $2.29 per GPU-hour and the H100 from $1.99 on-demand, verified 2026-08-22.
Which has more VRAM, GH200 or H100?
The GH200 has 96 GB and the H100 has 94 GB. The GH200 has more.
Prices render from the live dataset each build. More: the screener · price movers · methodology.