Video Generation

Generate videos from text prompts, a starting image, or reference images. The REST API is asynchronous: submit the job, then poll GET /v1/videos/{request_id} until it is done.

CLI

progrok video "a drone flying over mountains at sunrise" --duration 5 --resolution 480p
progrok video "animate this night sky" --model grok-imagine-video-1.5-preview --image sky.png
progrok video "openai-compatible duration alias" --seconds 5

API

# Submit generation request
curl -X POST http://localhost:18645/v1/videos/generations   -H 'Authorization: Bearer anything'   -H 'Content-Type: application/json'   -d '{
  "model": "grok-imagine-video",
  "prompt": "a drone flying over mountains at sunrise"
}'

# Poll for result
curl http://localhost:18645/v1/videos/vg_abc123   -H 'Authorization: Bearer anything'

Parameters

Parameter	Type	Description
prompt	string	Text description of the video
duration / seconds	integer	Length in seconds. `seconds` is the OpenAI-compatible alias.
aspect_ratio	string	16:9, 9:16, 1:1, 4:3, 3:4, 3:2, or 2:3
resolution	string	480p or 720p confirmed. Treat 1080p as conflicting until live smoke confirms it.
image	object	`{url \| file_id}` for image-to-video.
reference_images	array	`[{url \| file_id}]`, max 7, max 10 second output.
output.upload_url	string	Optional signed PUT destination for the result.

Mode Rules

REST has no mode body field. Text-to-video is prompt only, image-to-video adds image, and reference-to-video adds reference_images. Do not combine image and reference_images; xAI returns 400 Bad Request.

grok-imagine-video-1.5-preview is live-smoked as image-to-video only. Prompt-only T2V and reference_images currently return xAI 400 errors for that preview model.

Status Values

pending — Generation in progress.

done — Video ready for download.

failed — Generation failed.

expired — Download URL has expired.