H100
H100 is a modern high-tier tensor core processor. It's a good choice for training models or the best inference money can buy.
| 80 | 0.001097/s 3.9492/h -41% | ||||
| 80 | 0.001986/s 7.1496/h +7% | ||||
| 80 | 0.0025/s 9/h +34% |
| Vendor | Free Tier | True Serverless | Pricing Page | Deployment | Webhooks |
|---|---|---|---|---|---|
| Runpod | Unknown | true | https://www.runpod.io/serverless-gpu | Docker | Supported |
| Modal | 30 USD per month | true | https://modal.com/pricing | Modal SDK | Not supported |
| Beam.cloud | 10 hours | true | https://www.beam.cloud/pricing | Beam SDK | Supported |