Skip to main content
POST
seedance-2.5-ext Video Generation
seedance-2.5-ext is the promotional-channel model in the Seedance 2.5 family, supporting text-to-video and multimodal reference generation:
  • Text-to-video and reference-image generation
  • Combined generation with reference videos and reference audios (video editing, extension, and first/last-frame modes are not supported)
  • The API is asynchronous: submitting returns a generation.task; fetch results via the video task status endpoint.
Requests are routed to the optimal upstream automatically: image-only or text-only requests go to the low-cost upstream; requests with reference videos, reference audios, or more than 10 reference images go to the full-capability upstream.

Authentication

string
required
All requests require a Bearer Token. Get your key from the API Key management page and add it to the request header:

Body

string
default:"seedance-2.5-ext"
required
ToAPIs public model name. Always use seedance-2.5-ext.
string
Describe the scene, camera movement, subject actions, style, and sound atmosphere. When referencing assets, use tags like “image 1”, “video 1”, or “audio 1” in array order.
integer
Output video duration in seconds, from 10 to 30. The compatible field duration is also accepted; do not pass both.
15 seconds is not supported when the request carries reference videos, reference audios, or more than 10 reference images. Text-only requests and requests with at most 10 images are unaffected.
string
Output aspect ratio. Supported: 16:9, 9:16, 4:3, 3:4, 1:1. The compatible field aspect_ratio is also accepted.
string
default:"720p"
Output resolution. Fixed to 720p.
string
Public URL of a single reference image. For multiple images use reference_images.
string[]
Array of reference image URLs, up to 30. You can also use image_with_roles with role set to reference_image.
object[]
Array of reference videos, up to 10. Each item contains:
  • url: publicly accessible video URL
  • role: fixed to reference_video
object[]
Array of reference audios, up to 10. Each item contains:
  • url: publicly accessible audio URL
  • role: fixed to reference_audio
Audio cannot be the only input — the request must also include at least one reference image or reference video.
string
ToAPIs task completion callback URL. See Task Webhooks.
string
Your order number or business task ID. Stored with the task; you can also query it via GET /v1/videos/generations/{client_business_id}.
string
Custom trace ID for request tracking.

Media Combination Rules

Request Examples

Text-to-Video

Reference Image

Video and Audio Combined Reference

Response

string
Task ID for status queries.
string
Returned when provided in the request.
string
Fixed to generation.task.
string
The model used for this request.
string
queued, in_progress, completed, or failed.
integer
Task progress from 0 to 100.
integer
Creation timestamp.