About Replicate
Replicate is a platform that lets developers run AI models via an API, enabling one-line integration to build and scale AI-powered products. It serves developers and teams who want to run, fine-tune, and deploy models across image, speech, music, video, and language tasks.
Key Features
When you’re choosing a workflow for model use, having a single API that covers execution, customization, and hosting matters for moving from experiments to production:
One-line API integration
Run models in your apps with a single line of code, enabling rapid experimentation and quick onboarding.
Run and fine-tune models
Train models on your data, update them, and push new versions to production.
Custom deployment & hosting at scale
Deploy your own models or use hosted options with managed infrastructure, scaling as demand grows.
Multimodal model catalog
Access models across images, speech, music, video, and language tasks for versatile workflows.
Summary
Best for product and ML engineering teams building AI-powered apps, startups prototyping models, and data-ops teams deploying at scale.
Pricing
View pricingPublic Models
$0.015/ thousand output tokens / varies
- billed by time taken to run
- billed by input and output
Public Models Input
$3.00/ million input tokens / varies
- billed by time taken to run
- billed by input and output
black-forest-labs/flux-1.1-pro
$0.04/ output image / varies
- text-to-image model
- excellent image quality
black-forest-labs/flux-dev
$0.025/ output image / varies
- generates images from text descriptions
black-forest-labs/flux-schnell
$3.00/ thousand output images / varies
- fastest image generation model for local development
deepseek-ai/deepseek-r1
$0.01/ thousand output tokens / varies
- reasoning model trained with reinforcement learning
deepseek-ai/deepseek-r1 input
$3.75/ million input tokens / varies
- reasoning model trained with reinforcement learning
ideogram-ai/ideogram-v3-quality
$0.09/ output image / varies
- creates high-quality images
recraft-ai/recraft-v3
$0.04/ output image / varies
- text-to-image model
- generates long texts
wavespeedai/wan-2.1-i2v-480p
$0.09/ second of output video / varies
- accelerated inference for video generation
wavespeedai/wan-2.1-i2v-720p
$0.25/ second of output video / varies
- high resolution video generation
CPU (Small)
$0.000025/sec / $0.09/hr
- 2GB RAM, 1x CPU
CPU
$0.000100/sec / $0.36/hr
- 8GB RAM, 4x CPU
Nvidia A100 (80GB) GPU
$0.001400/sec / $5.04/hr
- 80GB GPU, 10x CPU
2x Nvidia A100 (80GB) GPU
$0.002800/sec / $10.08/hr
- 160GB GPU, 20x CPU
Nvidia H100 GPU
$0.001525/sec / $5.49/hr
- 80GB GPU, 13x CPU
Nvidia L40S GPU
$0.000975/sec / $3.51/hr
- 48GB GPU, 10x CPU
2x Nvidia L40S GPU
$0.001950/sec / $7.02/hr
- 96GB GPU, 20x CPU
Nvidia T4 GPU
$0.000225/sec / $0.81/hr
- 16GB GPU, 4x CPU
4x Nvidia A100 (80GB) GPU
$0.005600/sec / $20.16/hr
- Additional Multi-GPU A100 capacity with contracts
8x Nvidia A100 (80GB) GPU
$0.011200/sec / $40.32/hr
- Additional Multi-GPU A100 capacity with contracts
2x Nvidia H100 GPU
$0.003050/sec / $10.98/hr
- Additional Multi-GPU H100 capacity with contracts
4x Nvidia H100 GPU
$0.006100/sec / $21.96/hr
- Additional Multi-GPU H100 capacity with contracts
8x Nvidia H100 GPU
$0.012200/sec / $43.92/hr
- Additional Multi-GPU H100 capacity with contracts
4x Nvidia L40S GPU
$0.003900/sec / $14.04/hr
- Additional Multi-GPU L40S capacity with contracts
8x Nvidia L40S GPU
$0.007800/sec / $28.08/hr
- Additional Multi-GPU L40S capacity with contracts
