Generate videos using AI video models

🎬 AI Video Generation

Generate videos from a text prompt, or animate an existing image. Runs on the Vercel AI Gateway; the finished MP4 is stored on Blink's own CDN and returned as a permanent public URL safe to embed in a site.

Models are selected by tier, not by raw model ID:

fast (default) — cheapest and quickest, best for background/hero loops
better — higher fidelity, same controls
best — highest quality; noticeably slower, use only when asked for

Audio is off by default: web video autoplays muted, and muted generation is cheaper. Pass generate_audio: true to include a track.

Typical latency is 30–60s (fast/better) and up to ~150s (best).

post/api/ai/{project_id}/video

Request

Base URL
https://core.blink.new
URL
https://core.blink.new/api/ai/{project_id}/video
Auth
HTTP bearer

Path parameters

project_idstring required

Example:my-project-123

Unique identifier for the project

Request body

promptstring required

Description of the video, or of the motion to apply to image_url

model'fast' | 'better' | 'best'

Quality tier. Legacy fal-ai/* IDs are accepted and mapped to a tier.

image_urlstring uri

Source image to animate (image-to-video). Also accepts a data URI.

durationstring

Clip length in seconds; snapped to the nearest length the model supports.

aspect_ratio'16:9' | '9:16' | '1:1'

Text-to-video only. Ignored when image_url is set (output follows the image).

generate_audioboolean

Include an audio track. Off by default.

Example request

{
  "prompt": "A serene mountain lake at sunrise, mist rising from the water",
  "duration": "5s"
}

Response

Video generated

Changes

No recorded changes to this operation across all 1 revision of this API.