Replicate

Replicate is a cloud platform that lets you run thousands of open source AI models through a simple API — from image and video generation to speech and language models. It's built by a team backed by a16z, Sequoia and NVIDIA (Cloudflare acquired it in 2026) and is aimed mainly at developers and indie hackers who want an AI feature in their app without dealing with GPU servers or model training.

It works in a straightforward way: pick a model from the catalog, send it an input (text, image, audio) with a single API call, and get the output back. Replicate handles all the infrastructure behind the scenes — booting the model, GPUs, scaling under load. You only pay for the seconds your code actually runs; for public models you pay nothing for idle time.