GPU pricing for 39 clouds — AWS, Azure, Oracle, CoreWeave, Lambda, RunPod, Vast.ai and more — compared model by model on price. Every price below is per single GPU per hour, pulled from the provider's own pricing page or API and refreshed every 15 minutes.
| Provider | GPU models | Cheapest $/GPU-hr | Cheapest GPU | Offers |
|---|---|---|---|---|
| Vast.ai | 17 | $0.020 | V100 | 96 |
| Google Cloud | 6 | $0.056 | L4 | 12 |
| Verda | 11 | $0.085 | V100 | 86 |
| Salad Cloud | 2 | $0.160 | RTX 4090 | 2 |
| Runpod Community | 11 | $0.190 | V100 | 22 |
| Lium.io | 9 | $0.330 | L40 | 16 |
| GPU.ai | 6 | $0.340 | RTX 4090 | 6 |
| Novita | 4 | $0.340 | RTX 4090 | 7 |
| Thunder Compute | 4 | $0.350 | RTX A6000 | 4 |
| LeaderGPU | 5 | $0.380 | A40 | 5 |
| AWS | 6 | $0.423 | L40S | 18 |
| Runpod | 16 | $0.440 | A40 | 32 |
| Jarvis Labs | 3 | $0.440 | L4 | 3 |
| Seeweb | 8 | $0.442 | L4 | 8 |
| Hyperstack | 10 | $0.500 | RTX A6000 | 10 |
| Spheron | 10 | $0.530 | RTX 4090 | 10 |
| Prime Intellect | 11 | $0.540 | RTX A6000 | 11 |
| Massed Compute | 11 | $0.550 | RTX A6000 | 37 |
| Denvr Dataworks | 4 | $0.650 | A40 | 4 |
| Azure | 7 | $0.679 | A100 PCIe | 20 |
| Beam | 9 | $0.690 | RTX 4090 | 9 |
| Koyeb | 3 | $0.750 | RTX A6000 | 3 |
| Lambda Labs | 7 | $0.790 | V100 | 15 |
| Modal | 7 | $0.799 | L4 | 7 |
| Baseten | 4 | $0.848 | L4 | 4 |
| Cudo Compute | 3 | $0.870 | L40S | 3 |
| Scaleway | 5 | $0.917 | L4 | 16 |
| Nebius | 5 | $0.950 | RTX PRO 6000 | 10 |
| OVHcloud | 5 | $0.966 | L4 | 5 |
| fal | 5 | $1.10 | RTX PRO 6000 | 10 |
| CoreWeave | 8 | $1.25 | L40 | 8 |
| Civo | 4 | $1.29 | L40S | 12 |
| Crusoe | 6 | $1.50 | L40S | 6 |
| DigitalOcean | 9 | $1.57 | L40S | 15 |
| Voltage Park | 1 | $1.99 | H100 SXM | 1 |
| Oracle Cloud | 7 | $3.50 | L40S | 7 |
| Replicate | 4 | $3.51 | L40S | 16 |
| Together AI | 3 | $3.69 | H100 SXM | 6 |
| Latitude.sh | 1 | $24.00 | H100 SXM | 1 |