Provider pricing

Replicate AI API pricing

Replicate runs thousands of community and first-party models — Llama 4, FLUX, Stable Diffusion, Whisper, MusicGen — on a per-second GPU billing model. It's the easiest way to deploy a custom model: push a Cog container and Replicate handles autoscaling, queuing, and a public API.

Founded2019
HeadquartersSan Francisco, CA
Tracked modelsDirectory

Best for

Multimodal apps, custom model hosting, and teams that need more than just LLMs.

Strengths

  • Thousands of community models across text, image, video, audio
  • Per-second GPU billing — only pay while a request runs
  • Cog containers: deploy any custom model with a single push
  • Webhooks, streaming, and long-running prediction support out of the box