NVIDIA RTX 5090
32 GB GPU · from $0.69/hr · offered by Vast.ai
- VRAM
- 32 GB
- Architecture
- —
- Compute cap.
- —
- Mem. bandwidth
- —
- CUDA cores
- 21760
- Tensor cores
- 680
- TDP
- 575 W
- Interconnect
- PCIe
- Per hour
- $0.69
Price history
No observed price history for this listing yet. History appears as the catalog accumulates live observations.
Measured benchmarks
No measured benchmarks for this accelerator yet. Points appear as the catalog records observations from measured sources — GPUVerse never fabricates a benchmark.
Further reading
Sizing this accelerator for a model? The blog walks through the VRAM math and which tier fits each model size: How to choose a GPU for LLM inference.