What is a video generation API?

Updated 2026-10-17

A video generation API is a web service that produces video from instructions you send in code. You POST a prompt — optionally with a starting image, reference clips, or an existing video to edit — and get back a job. When the job completes, it contains a URL to the finished MP4.

The job lifecycle

Generation takes anywhere from several seconds to several minutes, so these APIs are asynchronous. A job moves through queued → in_progress → completed or failed. You poll the job (or use a webhook-style integration) rather than holding a connection open.

What you control

• Model — which checkpoint generates the video (Kling, Veo, Seedance, Wan, MiniMax…).
• Duration and resolution — billed per second of output, frequently at different rates per resolution tier.
• Input mode — text-to-video, image-to-video, reference-to-video, or video editing, depending on what the model supports.

Why one model has several prices

A model's creator usually has its own API, and third-party hosts resell the same checkpoint. Each host sets its own price and has its own reliability. A gateway like VideoRouter exposes all the hosts for a model behind one endpoint, so you can route to the cheapest healthy one or pin a specific provider.

How to choose

1. Pick the model for the look you need — the price table shows what each costs today.
2. Decide whether you need a particular host (data-handling, account relationship) or just the cheapest.
3. Prototype with short, low-resolution clips; only then switch to the premium tier.

Ready to try it? The quickstart has a working request in curl and Python.