RTX 3090 vs RTX 5090
RTX 3090 starts at $0.20/GPU-hour on-demand; RTX 5090 starts at $0.78/GPU-hour on-demand. Compare the exact variants, memory and published specifications below. GPU spec numbers are not benchmarks.
74.4% cheaper at RTX 3090.
updated 2026-10-11. Price alone does not establish equivalent capacity or performance.
RTX 3090
RTX 5090
Spec sheet
| RTX 3090 | RTX 5090 | |
|---|---|---|
| Spec variant | RTX 3090 | RTX 5090 |
| Architecture | Ampere | Blackwell |
| VRAM (spec variant) | 24 GB | 32 GB |
| FP16 TFLOPS (dense) | 142 | 210 |
| Memory bandwidth | 936 GB/s | 1792 GB/s |
| TDP | 350 W | 575 W |
| Interconnect | PCIe Gen4 | PCIe Gen5 |
| Launched | 2020 | 2025 |
Rental price per manufacturer-rated FP16 compute: RTX 3090 $1.41 per PFLOP-hour vs RTX 5090 $3.71. Formula: cheapest $/hr ÷ dense FP16 PFLOPS (1 PFLOP = 1000 TFLOPS). This uses RTX 3090 and RTX 5090, whose datasheets may differ from the variants above. It is not measured job throughput. Specs: RTX 3090 manufacturer, RTX 5090 manufacturer.
What they cost to rent (2026-10-11)
| RTX 3090 | RTX 5090 | |
|---|---|---|
| Cheapest on-demand $/hr | $0.20 | $0.78 |
| Cheapest spot/community $/hr | $0.09 | $0.25 |
| Cheapest source | TensorDock | Spheron |
| Providers offering it | 4 | 5 |
Cheapest on-demand rate; the cheaper one is highlighted. Spot/marketplace tiers shown on the individual pages.
FAQ
RTX 3090 is cheaper: the RTX 3090 rents from $0.20 per GPU-hour and the RTX 5090 from $0.78 on-demand, updated 2026-10-11.
The largest recorded RTX 3090 variant has 24 GB and the largest RTX 5090 variant has 32 GB. Other variants may have less memory. The RTX 5090 has more.
Compare more rental options: the screener · price movers · methodology.