Skip to main content
Use text_to_video for prompt-only generation or image_to_video with an owner-scoped source_creation_id from the generated image library.
If audio is enabled, use audio_prompt to describe the intended sound. Query the model catalog with the exact video mode before submitting. It returns only the selections currently available for that workflow.
Video generation can take longer than image generation. Prefer signed webhooks for production workflows and retain polling as a recovery path.