L4 vs RTX 5090
The cheapest L4 ($0.083) costs 62% less per hour than the cheapest RTX 5090 ($0.22).
Per unit of performance, the RTX 5090 is 8% cheaper: $0.24 versus $0.26 per A100-equivalent hour.
The RTX 5090 has 32 GB of memory against 24 GB on the L4.
| Measure | NVIDIA L4 24GB | NVIDIA GeForce RTX 5090 |
|---|---|---|
| Lowest per GPU-hour | $0.083 | $0.22 |
| Median per GPU-hour | $0.80 | $0.87 |
| On-demand from | $0.10 | $0.34 |
| Spot from | $0.083 | $0.25 |
| Reserved from | $0.25 | $0.22 |
| Lowest per A100-equivalent hour | $0.26 | $0.24 |
| Live offers | 156 | 405 |
| Providers | 50 | 84 |
| Memory | 24 GB GDDR6 | 32 GB GDDR7 |
| Memory bandwidth | 300 GB/s | 1,792 GB/s |
| FP16 / BF16 | 121 TFLOPS | 209.5 TFLOPS |
| FP8 | 242.5 TFLOPS | 419 TFLOPS |
| Performance vs A100 80GB | 0.32× | 0.92× |
| Architecture | Ada Lovelace | Blackwell |
| Power | 72 W | 575 W |