Replicate logo

Replicate πŸ‡ΊπŸ‡Έ

Run AI/ML models without complex infrastructure

Founded in 2019, Replicate is a platform to train and deploy ML models in the cloud. You can run and fine tune open source models such as Meta's Llama 3, Mistral and Stable Diffusion without having to set up complex infrastructure.

They also feature a collection of generative models shared by the community, which you can run via the web UI or using their REST API.

Example customers include BuzzFeed, PhotoAI, Magnific, Unsplash, HeadshotPro.

What's good about Replicate

  • Run AI/ML models without worrying about infrastructure and scaling
  • Excellent web UI to explore and try out models without code
  • Train and deploy custom models using Cog

What does Replicate do?

Here are some of the managed services that Replicate offers:

Replicate's Pricing

Replicate uses a metered billing model. You only pay as long as your code is running, billed by the second based on the hardware selected.

Here's a cost estimate for the various GPU models available:

Name Total GPU Memory vCPUs RAM Price (hour)
Nvidia T4 16GB 4 16GB $0.81
Nvidia A40 (Small) 48GB 4 16GB $2.07
Nvidia A40 (Large) 48GB 10 72GB $2.61
Nvidia A100 (40GB) 40GB 10 72GB $4.14
Nvidia A100 (80GB) 80GB 10 144GB $5.04
2x Nvidia A40 (Large) 96GB 20 144GB $5.22
2x Nvidia A100 (40GB) 80GB 20 144GB $8.28
2x Nvidia A100 (80GB) 160GB 20 288GB $10.08
4x Nvidia A40 (Large) 192GB 40 288GB $10.44
4x Nvidia A100 (40GB) 160GB 40 288GB $16.56
4x Nvidia A100 (80GB) 320GB 40 576GB $20.16
8x Nvidia A40 (Large) 384GB 48 680GB $20.88
8x Nvidia A100 (80GB) 640GB 80 960GB $40.32

You can find Replicate's latest pricing here.

Note: Our pricing examples are based on several assumptions. Your actual costs may differ. Always check the cloud provider's website for the most up-to-date pricing.

Replicate Alternatives

Here are some alternatives to Replicate:

Our data for Replicate was last updated on June 3, 2024.