H100 vs RTX 4090
Datasheet specs side by side, plus what each card actually costs to rent today across every provider we track.
H100
RTX 4090
Spec sheet
| H100 | RTX 4090 | |
|---|---|---|
| Architecture | Hopper | Ada Lovelace |
| VRAM (max variant) | 94 GB | 24 GB |
| FP16 TFLOPS (dense) | 989 | 165 |
| Memory bandwidth | 3350 GB/s | 1008 GB/s |
| TDP | 700 W | 450 W |
| Interconnect | NVLink 900GB/s | PCIe Gen4 |
| Launched | 2022 | 2022 |
Computed price-performance at today's cheapest on-demand rates: H100 $2.01 per PFLOP-hour vs RTX 4090 $2.12. Formula: cheapest $/hr ÷ dense FP16 PFLOPS (1 PFLOP = 1000 TFLOPS). Specs from vendor datasheets (source, source).
What they cost to rent (2026-08-22)
| H100 | RTX 4090 | |
|---|---|---|
| Cheapest on-demand $/hr | $1.99 | $0.35 |
| Cheapest spot/community $/hr | $0.94 | $0.16 |
| Cheapest source | Voltage Park | TensorDock |
| Providers offering it | 32 | 4 |
Cheapest verified on-demand rate per side from today's snapshot; highlighted = cheaper. Spot/marketplace tiers shown on the individual pages.
FAQ
RTX 4090 is cheaper: the H100 rents from $1.99 per GPU-hour and the RTX 4090 from $0.35 on-demand, verified 2026-08-22.
The H100 has 94 GB and the RTX 4090 has 24 GB. The H100 has more.
Prices render from the live dataset each build. More: the screener · price movers · methodology.