ドキュメントGrok Video
Model
grok-video
Dynamic-duration video model that needs referenceTaskId for image-to-video.
Model supplement for /unified-video/quote and /unified-video/generate権限: api.generate
向いている用途
Teams that want a dynamic quote model with text and image modes.
Text-to-video
Image-to-video with referenceTaskId
Dynamic duration pricing
What you can send
- Required: model, prompt
- Optional: images (0-2), referenceTaskId, aspectRatio, duration, resolution, mode, projectId
- If images are present, referenceTaskId becomes required
Field rules
duration
Supported integer range: 6..30.
resolution
Supported: 480p, 720p.
mode
Default normal.
referenceTaskId
Required for image-to-video. Forwarded upstream as task_id.
Pricing formula
Resolution
Formula
480p
9 + (duration - 6) * 2
720p
17 + (duration - 6) * 2
Frontend notes
- Show referenceTaskId only when image mode is active.
- Allow any integer duration from 6 to 30.
- Use callback as the main completion flow.
すぐに試したいですか？
Settings で general API key を作成してから、ここに戻ってリクエスト例をコピーしてください。