GPU Cloud Providers
Compare GPU cloud providers for AI training and inference. Live pricing data updated hourly from public APIs and vendor documentation.
All Providers
| PROVIDER | GPUS AVAILABLE | STARTING PRICE | KEY STRENGTHS |
|---|---|---|---|
| Azure | A100, H100, H200 | $1.50–$4.50/hr | Enterprise SLAs, global regions, hybrid cloud |
| CoreWeave | A100, H100, H200 | $2.50–$5.50/hr | GPU-native, high availability, Kubernetes |
| Cudo Compute | A100, RTX 4090 | $0.50–$2.50/hr | Sustainable, distributed, competitive pricing |
| FluidStack | A100, H100, L40S | $1.00–$3.50/hr | Low-cost, on-demand, API-first |
| Lambda | A100, H100, H200 | $1.25–$3.99/hr | ML-optimized, Lambda Stack, reserved capacity |
| Nebius | H100, H200, L40S | $1.18–$5.50/hr | EU data sovereignty, DGX-ready, ML platform |
| Oracle Cloud (OCI) | A100, H100, H200 | $2.00–$5.00/hr | Bare metal, RDMA networking, superclusters |
| Paperspace | A100, H100, RTX 4090 | $0.80–$3.09/hr | Gradient platform, notebooks, easy onboarding |
| RunPod | A100, H100, RTX 4090 | $0.44–$3.89/hr | Serverless GPU, spot pricing, community cloud |
| Together AI | A100, H100 | $1.00–$4.00/hr | Inference-optimized, open-source models |
| Vast.ai | A100, RTX 4090, L40S | $0.30–$2.80/hr | Marketplace model, cheapest spot GPUs |
| Vultr | A100, H100, L40S | $0.90–$3.50/hr | Global edge, simple API, bare metal |
Compare live GPU pricing across all major providers. View full comparison →