Creates a video generation job from a text prompt. This is an asynchronous operation that returns immediately with a job ID. Use the retrieve endpoint to check the status and get the video URL when generation is complete.
Model identifier in format provider/model
Text prompt describing the video to generate
Optional reference image for image-to-video. OpenAI and Gemini require a base64 data URL (e.g., data:image/png;base64,...). Runway and Replicate accept both data URLs and plain URLs.
Duration of the video in seconds as a string (e.g., "4")
Resolution of the generated video (e.g., 1280x720, 720x1280, 1920x1080)
Text describing what to avoid in the generated video
Seed for reproducible generation
Source video URI for video-to-video generation (provider-specific, e.g. GCS URI)
Enable audio generation in the video (supported by select providers/models)
Fallback models to try if primary model fails
Successful response. Returns a video generation job object with status information. Poll the retrieve endpoint to check completion status.
Provider-native job ID. To use in path parameters (retrieve/delete/download), combine as {id}:{provider} (e.g., task_abc123:runway)
Object type, always "video"
video Model used for video generation
Current lifecycle status of the video generation job:
queued: Job is waiting to be processedin_progress: Video is currently being generatedcompleted: Video generation completed successfullyfailed: Video generation failedqueued, in_progress, completed, failed Approximate completion percentage (0-100)
0 <= x <= 100Prompt used to generate the video
Source video ID if this is a remix
Duration of the generated video in seconds as a string (e.g., "4")
Resolution of the generated video (e.g., "1920x1080")
Unix timestamp (seconds) when the job was created
Unix timestamp (seconds) when the job completed
Unix timestamp (seconds) when downloadable assets expire
Generated video outputs (only present when status is "completed")
Information about content that was filtered due to safety policies
Additional fields included in responses