Models and inputs
Choose a public Seedance model. The API selects the workflow from the media URLs you provide.
Use one of the public model IDs below. Open AI Image keeps provider model names internal, so they can change without affecting the public API.
| Model | Typical use | References | Duration | Resolution |
|---|---|---|---|---|
seedance_2_5 | Text, frame, and reference generation | 50 | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 s | 480p, 720p, 1080p |
seedance_2_0 | High-quality multimodal generation | 15 | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 s | 480p, 720p, 1080p, 4K |
seedance_2_0_fast | Faster Seedance 2.0 generation | 15 | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 s | 480p, 720p |
seedance_2_0_mini | Seedance 2.0 with lower credit use | 15 | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 s | 480p, 720p |
The API checks the selected model and parameters before creating a task. Unsupported combinations return 400.
Automatic generation mode
Use a public model ID and provide reference media in reference_urls. With no media, the API creates a video from text. Supported models can use image or video references.
Reference media must be uploaded to your account first. Use the owner-scoped /api/studio/media/ URL returned by the upload endpoint; arbitrary remote URLs and legacy /uploads/ paths are rejected. Supported media types and counts depend on the selected model.
Parameters
Set parameters with the selected model’s duration, resolution, and aspect_ratio. Enable generate_audio only when the model supports synchronized audio.
Seedance 2.5 media limits
- Upload files to your account and pass the returned private media URLs rather than Base64 or arbitrary external URLs.
- Keep the reference count within the selected model’s limit.
- Keep each JSON request body within 2 MB.