Replicate GPU Cloud Pricing
from $0.810/hrServerless platform for running and hosting machine learning models via API, billed per second of GPU compute with a large community model library.
Start on Replicate
Current Pricing — 13 configurations
Some links are affiliate links — we may earn a commission at no cost to you.Details.
| Provider | Price/hr | Configuration | Region | Billing | Availability | Rent |
|---|---|---|---|---|---|---|
| Replicatecheapest | $0.810/hr | 1x Nvidia T4 GPU gpu-t4 | us-east | per-second | on-demand | Rent → |
| Replicate | $3.51/hr | 1x Nvidia L40S GPU gpu-l40s | us-east | per-second | on-demand | Rent → |
| Replicate | $5.04/hr | 1x Nvidia A100 (80GB) GPU gpu-a100-large | us-east | per-second | on-demand | Rent → |
| Replicate | $5.49/hr | 1x Nvidia H200 GPU gpu-h200 | us-east | per-second | on-demand | Rent → |
| Replicate | $7.02/hr | 1x 2x Nvidia L40S GPU gpu-l40s-2x | us-east | per-second | on-demand | Rent → |
| Replicate | $10.08/hr | 1x 2x Nvidia A100 (80GB) GPU gpu-a100-large-2x | us-east | per-second | on-demand | Rent → |
| Replicate | $10.98/hr | 1x 2x Nvidia H200 GPU gpu-h200-2x | us-east | per-second | on-demand | Rent → |
| Replicate | $14.04/hr | 1x 4x Nvidia L40S GPU gpu-l40s-4x | us-east | per-second | on-demand | Rent → |
| Replicate | $20.16/hr | 1x 4x Nvidia A100 (80GB) GPU gpu-a100-large-4x | us-east | per-second | on-demand | Rent → |
| Replicate | $21.96/hr | 1x 4x Nvidia H200 GPU gpu-h200-4x | us-east | per-second | on-demand | Rent → |
| Replicate | $28.08/hr | 1x 8x Nvidia L40S GPU gpu-l40s-8x | us-east | per-second | on-demand | Rent → |
| Replicate | $40.32/hr | 1x 8x Nvidia A100 (80GB) GPU gpu-a100-large-8x | us-east | per-second | on-demand | Rent → |
| Replicate | $43.92/hr | 1x 8x Nvidia H200 GPU gpu-h200-8x | us-east | per-second | on-demand | Rent → |
Provider Details
FAQ
What GPUs does Replicate offer?
1x Nvidia A100 (80GB) GPU
gpu-a100-large, 1x 2x Nvidia A100 (80GB) GPU
gpu-a100-large-2x, 1x Nvidia L40S GPU
gpu-l40s, 1x 2x Nvidia L40S GPU
gpu-l40s-2x, 1x Nvidia T4 GPU
gpu-t4, 1x 4x Nvidia A100 (80GB) GPU
gpu-a100-large-4x, 1x 8x Nvidia A100 (80GB) GPU
gpu-a100-large-8x, 1x Nvidia H200 GPU
gpu-h200, 1x 2x Nvidia H200 GPU
gpu-h200-2x, 1x 4x Nvidia H200 GPU
gpu-h200-4x, 1x 8x Nvidia H200 GPU
gpu-h200-8x, 1x 4x Nvidia L40S GPU
gpu-l40s-4x, 1x 8x Nvidia L40S GPU
gpu-l40s-8x
Where are Replicate data centers located?
Replicate operates in: us-central.
How does Replicate bill for GPU usage?
Replicate supports per-second billing.
Is Replicate reliable for production workloads?
Replicate has a trust score of 4.3/5. Features include: serverless, model-hosting, api-access, auto-scaling.
Last data refresh: October 19, 2026. Verify on Replicate's site.
Related Providers
Fal.ai
Serverless inference platform optimized for generative media workloads (images, video, audio) with sub-second cold starts and real-time streaming.
Modal
Developer-first serverless GPU platform with a Python-native SDK, per-second billing, and automatic cold-start optimization for ML workloads.
RunPod
Consumer and data-center GPU cloud with spot and on-demand instances, a large template marketplace, and a serverless inference platform.
TensorDock
Budget-friendly GPU marketplace aggregating data-center hardware from multiple hosts, offering broad GPU variety at competitive hourly rates.
Genesis Cloud
European GPU cloud powered by renewable energy, offering NVIDIA instances at competitive rates with a focus on sustainability.
Vast.ai
Peer-to-peer GPU marketplace that aggregates idle hardware from independent hosts, offering some of the lowest per-hour rates available.