Replicate
Run open-source AI models via API with per-second or per-token billing across GPUs
Replicate doesn't publish a pricing.md file yet.
This pricing data is community-maintained and last verified 2026-10-11.
Pricing Tiers
| Tier | Price | Includes | Overage |
|---|---|---|---|
|
Free
free
|
Free
|
— | — |
|
Pay As You Go
usage based
|
Free
|
— |
gpu-compute-A100-80GB: $0.0032 per seconds
gpu-compute-CPU: $0.000025 per seconds
|
Portability
Switching Cost
moderate
Open Standard
Proprietary
What You Lose
Replicate-specific prediction API format, custom model deployments, fine-tuned model hosting
Transparency Score
B
60/100
Pricing Transparency Score
Score breakdown
✕
Publishes pricing.md
0/30
✓
Free tier available
10/10
✓
All prices public
15/15
✓
Usage pricing documented
15/15
✓
Portability documented
10/10
✕
Open standard
0/10
✓
Recently verified
10/10