Replicate
replicate.com · Inference & Hosting
Run and deploy open models through a simple API, packaging custom models with Cog.
Replicate runs machine learning models through a simple HTTP API: thousands of public community models plus hosted proprietary ones, covering image, video, audio and language models. You can also package and deploy your own model with Cog, Replicate's open source container tool.
Pricing, per the official pricing page: you pay only for what you use. Most models are billed per second of runtime, at a rate that depends on the hardware, and some are billed per input and output (per image or per token); each model page shows an estimate. In the Vercel AI SDK, @ai-sdk/replicate covers Replicate's image and video models. For high volume LLM serving, dedicated inference platforms such as Baseten, Together AI or Fireworks AI are the usual alternatives.
Replicate is one of 10 inference and hosting tools on Drydock. See all inference and hosting tools or how we check facts.
