Skip to main content
POST

Grok Imagine Video

grok-imagine-video is the video generation model of the Grok Imagine series. It uses a JSON request body and supports text-to-video and reference-image video generation.
  • The endpoint path is POST /v1/videos.
  • The request format is application/json.
  • Text-to-video only requires model, prompt, seconds, aspect_ratio, and resolution.
  • For image-to-video, use image for a single image and images for multiple images.
  • resolution supports 480P and 720P.
  • prompt supports up to 4096 characters.

Method and Path

Response Example

Authentication

Body

model
string
required
Always pass grok-imagine-video.
prompt
string
required
Video generation prompt. Up to 4096 characters.
seconds
string
Target duration in seconds. Minimum is 1 second; passing it as a string is recommended, e.g. "6".
aspect_ratio
string
Aspect ratio. Supports common presets 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 2:1, 1:2, 19.5:9, 9:19.5, 20:9, 9:20, as well as custom number:number ratios.
resolution
string
Output resolution. Supports 480P and 720P; numeric forms like 480 and 720 are also accepted.
image
string
Single reference image, as a data URI such as data:image/png;base64,....
images
array<string>
Multiple reference images. Each array member is a data URI such as data:image/png;base64,.... Do not pass image together with images.

Response

id
string
Task ID. Query the result later with GET /v1/videos/{id}.
object
string
Object type, usually video.
model
string
The model name actually submitted.
status
string
Task status. Common values are queued, processing, completed, failed, cancelled.
progress
integer
Task progress percentage.
video_url
string
Video URL after the task completes. You can also download the result with GET /v1/videos/{task_id}/content.

Differences from Grok 1.5 / 3