Replicate alternatives
Every model hosting and inference provider this site tracks, ordered by iScore — the same peers Replicate's own score is measured against. The line under each name is where its sourced facts differ from Replicate's; where a full comparison exists, the row links to it.
- RunPodPricing model: Prepaid credit balance drawn down by usage: all compute and storage billed per second, wi…
- Modalvs ReplicatePricing model: A platform fee plus metered compute, and the fee buys a compute allowance rather than acc…
- BasetenPricing model: No platform fee: per-token Model APIs for open models, and dedicated deployments billed p…
Where Replicate stands against these 3
Replicate's own sourced figures, placed among the alternatives above that document the same thing in the same unit. A rank here is one criterion, not a verdict; the iScore above weighs all of them.
- H100 SXM, per GPU-hour
- $5.49 per hour for a single Nvidia H100 with 80GB of GPU RAM, published as $0.001525 per second — the rate…3rd cheapest of the 4 that document it. Cheapest: Modal — $3.95 per hour for an Nvidia H100 SXM5, published as $0.001097 per second.
- Cheapest GPU, per hour
- $0.81 per hour for an Nvidia T4 with 16GB, the cheapest GPU on the hardware list3rd cheapest of the 3 that document it. Cheapest: RunPod — $0.27 per hour for an RTX A5000 with 24GB, on Secure Cloud (the page's default view).
- Largest GPU offered
- 160GB on the 2x Nvidia A100 (80GB) instance, the largest GPU RAM figure the hardware table states3rd highest of the 3 that document it. Highest: RunPod — 288GB per GPU on B300; B200 at 180GB is $7.99 an hour and H200 at 141GB is $5.29, on Secure Cloud (the page's…
A rank whose 80% interval overlaps its neighbour's is an ordering, not a finding; the intervals are on each provider's page and in /scores.json.