← GPU cloud (on-demand)
Compare 4 gpu cloud (on-demand) APIs
| EC2 p5 (H100 SXM ×8), per GPU Amazon Web Services | DeepInfra dedicated A100 80GB DeepInfra | Nebius L40S (Intel CPU) Nebius | DigitalOcean GPU Droplet L40S DigitalOcean (Paperspace) | |
|---|---|---|---|---|
| $ / GPU-hr | $6.88 Unverified sample | $0.89 Verified Oct 6, 2026 | $1.55 Verified Oct 6, 2026 | $1.57 Verified Oct 6, 2026 |
| Quality | 90 | – | – | – |
| Speed | 85 | – | – | – |
| GPU | H100 SXM | A100 80GB | L40S | L40S |
| VRAM | 80 GB | 80 GB | 48 GB | 48 GB |
| Interconnect | infiniband | — | none | — |
| Regions | us, eu, asia | — | — | — |
| Per-second billing | Yes | — | — | — |