Replicate lists 4 GPU models across 16 configurations, starting at $3.51 per GPU-hour for the L40S. On every GPU it offers, at least one of the 34 other dedicated providers we track is currently cheaper — the table below shows which, and by how much.
| GPU | VRAM | Pricing | Replicate $/GPU-hr | Best dedicated | Difference | Cheapest at | |
|---|---|---|---|---|---|---|---|
| L40S | 48 GB | on-demand | $3.51 | $0.550 | +538% | Novita | rent |
| L40S | 48 GB | reserved | $3.51 | $0.655 | +435% | LeaderGPU | rent |
| A100 SXM | 80 GB | on-demand | $5.04 | $1.09 | +362% | Thunder Compute | rent |
| A100 SXM | 80 GB | reserved | $5.04 | $0.995 | +406% | LeaderGPU | rent |
| H100 SXM | 80 GB | on-demand | $5.49 | $1.79 | +207% | Cudo Compute | rent |
| H100 SXM | 80 GB | reserved | $5.49 | $1.89 | +190% | fal | rent |
| H200 | 141 GB | reserved | $5.49 | $2.10 | +161% | fal | rent |
How much does a GPU cost on Replicate?
Replicate starts at $3.51 per GPU-hour for the L40S. It lists 4 GPU models in total: A100 SXM, H100 SXM, H200, L40S.
Is Replicate cheaper than other GPU clouds?
Not at the moment. For every GPU and pricing model Replicate offers, at least one of the 34 other dedicated providers we track is currently cheaper; the table above names which one and the size of the gap. Comparisons are like-for-like on GPU, pricing model and capacity tier.
Which GPUs does Replicate offer?
4 models: A100 SXM, H100 SXM, H200, L40S. Availability changes; the "in stock" flags on each GPU page reflect the last scrape.