SeeDance Video API Documentation

ByteDance Doubao SeeDance provides high-quality video generation models. This document describes the complete API interface specification for using SeeDance 2.0 and SeeDance 2.5 models for video generation. All video generation calls use the same /v1/video/generations endpoint, with different parameters depending on the use case.

Supported Models

Currently supported models include:

Model	Description
doubao-seedance-2-0-fast-260128	SeeDance 2.0 fast video generation model
doubao-seedance-2-0-260128	SeeDance 2.0 video generation model
doubao-seedance-2-5-260628	SeeDance 2.5 video generation model
dreamina-seedance-2-0-fast-260128	SeeDance 2.0 fast video generation model (overseas version)
dreamina-seedance-2-0-260128	SeeDance 2.0 video generation model (overseas version)
dreamina-seedance-2-5-260628	SeeDance 2.5 video generation model (overseas version)

The request format is identical for every model above. The allowed values for some parameters differ between the 2.0 and 2.5 families — see Model Differences.

Overview

The SeeDance video generation feature provides an asynchronous task processing mechanism:

Submit Task: Send an image and text prompt to create a video generation task
Query Status: Query generation progress and status through task ID
Get Results: Retrieve the generated video file after task completion

Task Status Flow

queued → in_progress → completed
                ↓
            failed

queued: Task has been submitted and is waiting to be processed
in_progress: Task is being processed
completed: Task completed successfully, video has been generated
failed: Task failed

API List

Method	Path	Description
POST	/v1/video/generations	Submit video generation task
GET	/v1/video/generations/{task_id}	Query task status

Model Differences

SeeDance 2.5 accepts the same request shape as 2.0, but raises every reference-input ceiling, adds audio-only input and mov output, and constrains the aspect ratio in some modes. Requests that exceed the limits of the selected model are rejected with a 400 invalid_request error before they reach the upstream provider.

Parameter / capability	SeeDance 2.0 series	SeeDance 2.5
`duration`	Default `5`. Integers `4`–`15`, or `-1` for automatic duration	Default `-1`. Integers `4`–`30`, or `-1` for automatic duration. Video editing tasks accept `-1` only
`size` (resolution)	Default `720p`. `480p`, `720p`; `doubao-seedance-2-0-260128` and `dreamina-seedance-2-0-260128` also accept `1080p` and `4k`	Default `720p`. `480p`, `720p`
`metadata.ratio`	Default `adaptive`. `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, `adaptive`	Same values for text-to-video and reference-to-video. First/last-frame, video editing and video extension accept `adaptive` only
Reference images	Up to 9	Up to 30
Reference videos	Up to 3; see constraints	Up to 10; see constraints
Reference audio	Up to 3; requires at least one reference image or video	Up to 10; may be used on its own (see Audio-Only Reference)
Image input formats	`jpeg`, `png`, `webp`, `bmp`, `tiff`, `gif`	Also `heic` and `heif`
`metadata.output_format`	Not supported (always `mp4`)	`mp4` (default) or `mov`
Video editing / extension	Not supported	Supported (see Video Editing and Extension)

Additional `metadata` fields

These optional fields are forwarded to the model as-is and apply to all SeeDance models unless noted otherwise.

Field	Type	Description
metadata.resolution	string	Alternative to the top-level `size` field; overrides it when both are present
metadata.duration	integer	Alternative to the top-level `duration` field; overrides it when both are present
metadata.generate_audio	boolean	Default `true`. When `true` the model generates voice, sound effects and background music matching the footage; `false` produces a silent video
metadata.return_last_frame	boolean	Default `false`. When `true` the last frame of the generated video is also returned, as a PNG with the same pixel dimensions and no watermark
metadata.watermark	boolean	Default `false`. When `true` an `AI Generated` watermark is shown in the lower-right corner
metadata.output_format	string	SeeDance 2.5 only. `mp4` (default) or `mov`. `mov` uses high-colour-precision professional encoding (H.264 / YUV 4:4:4 / PCM audio) that some players cannot open; it is recommended for video editing and extension workflows
metadata.priority	integer	Default `0`. Range `0`–`9`; a higher value moves the task ahead of lower-priority queued tasks
metadata.execution_expires_after	integer	Default `172800` (48 hours). Range `3600`–`259200` seconds. A task still queued or running after this time is terminated
metadata.safety_identifier	string	Stable, opaque per-end-user identifier, at most 64 characters. Use a hash of the user ID rather than personal data
metadata.tools	array of object	Opts into upstream tooling, e.g. `[{"type": "web_search"}]`. SeeDance 2.0 and 2.5 only
metadata.callback_url	string	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

Metadata keys other than the ones documented here (and imageMode, referenceVideos, referenceAudio, which are consumed while building the request) are ignored.

Usage Examples

Examples 1–5 apply to both the 2.0 and 2.5 families; substitute any supported model name and observe the limits in Model Differences. Examples 6–7 cover capabilities that only exist on SeeDance 2.5.

1. Basic Text Prompt

Text-to-video uses prompt and model. Optional top-level fields include duration and size. Other options can go in metadata as needed.

Request body:

{
    "prompt": "A large hand-shaped marble statue with vines growing on it, on a platform in the middle of a shallow pool of water.",
    "model": "doubao-seedance-2-0-fast-260128",
    "duration": 8,
    "size": "720p",
    "metadata": {
        "ratio": "16:9",
        "callback_url": ""
    }
}

Field	Type	Required	Description
prompt	string	Yes	Text prompt for the video
model	string	Yes	Model name (e.g. `doubao-seedance-2-0-fast-260128`)
duration	integer	No	Default `5` (2.0) or `-1` (2.5). Allowed: integers from `4` to `15` (2.0) or `4` to `30` (2.5), inclusive, or `-1` for automatic duration selection
size	string	No	Default `720p`. Allowed: `480p`, `720p`; the non-fast 2.0 models also accept `1080p` and `4k`
metadata	object	No	Extra metadata
metadata.ratio	string	No	Default `adaptive`. Allowed: `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, `adaptive`
metadata.callback_url	string	No	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

2. First/Last Frame Images

Provide one or two image URLs in images: the first entry is the first frame; when two URLs are given, the second is the last frame. Optional top-level duration and size match Basic Text Prompt.

Request body:

{
    "prompt": "The girl in the frame says cheese to the camera; 360-degree orbit camera move.",
    "model": "doubao-seedance-2-0-fast-260128",
    "images": [
        "https://ark-project.tos-cn-beijing.volces.com/doc_image/seepro_first_frame.jpeg",
        "https://ark-project.tos-cn-beijing.volces.com/doc_image/seepro_last_frame.jpeg"
    ],
    "duration": 8,
    "size": "720p",
    "metadata": {
        "ratio": "16:9",
        "callback_url": ""        
    }
}

Field	Type	Required	Description
prompt	string	Yes	Text prompt for the video
model	string	Yes	Model name (e.g. `doubao-seedance-2-0-fast-260128`)
images	array of string	Yes	One or two image URLs: first frame, and optionally last frame
duration	integer	No	Default `5` (2.0) or `-1` (2.5). Allowed: integers from `4` to `15` (2.0) or `4` to `30` (2.5), inclusive, or `-1` for automatic duration selection
size	string	No	Default `720p`. Allowed: `480p`, `720p`; the non-fast 2.0 models also accept `1080p` and `4k`
metadata	object	No	Extra metadata
metadata.ratio	string	No	Default `adaptive`. Allowed: `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, `adaptive`. SeeDance 2.5 accepts `adaptive` only in this mode — the output always keeps the aspect ratio of the first frame
metadata.callback_url	string	No	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

3. Reference Images

This flow also uses the top-level images array. Set metadata.imageMode to reference so each URL is treated as a reference image (not first/last frame). In this mode you can pass up to nine images on 2.0 models and up to thirty on 2.5 models. Optional top-level duration and size work the same as in Basic Text Prompt. In the prompt, reference images with tags such as [image 1], [image 2], … in the same order as images.

Request body:

{
    "prompt": "[image 1] A boy wearing glasses and a blue T-shirt and [image 2] a corgi puppy, sitting on the lawn in [image 3], 3D cartoon style.",
    "model": "doubao-seedance-2-0-fast-260128",
    "images": [
        "https://ark-project.tos-cn-beijing.volces.com/doc_image/seelite_ref_1.png",
        "https://ark-project.tos-cn-beijing.volces.com/doc_image/seelite_ref_2.png",
        "https://ark-project.tos-cn-beijing.volces.com/doc_image/seelite_ref_3.png"
    ],
    "duration": 8,
    "size": "720p",
    "metadata": {
        "imageMode": "reference",
        "ratio": "16:9",
        "callback_url": ""
    }
}

Field	Type	Required	Description
prompt	string	Yes	Text prompt; use `[image 1]` … `[image N]` to refer to images in order
model	string	Yes	Model name (e.g. `doubao-seedance-2-0-fast-260128`)
images	array of string	Yes	Reference image URLs (up to 9 on 2.0, 30 on 2.5)
duration	integer	No	Default `5` (2.0) or `-1` (2.5). Allowed: integers from `4` to `15` (2.0) or `4` to `30` (2.5), inclusive, or `-1` for automatic duration selection
size	string	No	Default `720p`. Allowed: `480p`, `720p`; the non-fast 2.0 models also accept `1080p` and `4k`
metadata	object	No	Must include `imageMode`: `reference` for reference-image mode
metadata.imageMode	string	Yes (this mode)	Set to `reference`
metadata.ratio	string	No	Default `adaptive`. Allowed: `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, `adaptive`
metadata.callback_url	string	No	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

Constraints on each reference image:

Count: up to 9 images on 2.0 models, up to 30 on 2.5 models.
Format: jpeg, png, webp, bmp, tiff, gif; 2.5 models also accept heic and heif.
Aspect ratio (width ÷ height): between 0.4 and 2.5 (inclusive).
Width and height (pixels): each between 300 and 6000 (inclusive).
File size: each image must not exceed 30 MB, and the whole request body must not exceed 64 MB. Prefer URLs over base64 for large files.

4. Reference Videos

Pass reference clip URLs in metadata.referenceVideos (up to three on 2.0 models, up to ten on 2.5 models). The prompt can refer to them in order (e.g. [video 1], [video 2], [video 3]).

Request body:

{
    "prompt": "In [video 1], the arched window opens and we enter the gallery interior; cut to [video 2], then the camera moves into the painting; cut to [video 3].",
    "model": "doubao-seedance-2-0-fast-260128",
    "size": "720p",
    "duration": 8,
    "metadata": {
        "ratio": "16:9",
        "referenceVideos": [
            "https://ark-project.tos-cn-beijing.volces.com/doc_video/r2v_extend_video1.mp4",
            "https://ark-project.tos-cn-beijing.volces.com/doc_video/r2v_extend_video2.mp4",
            "https://ark-project.tos-cn-beijing.volces.com/doc_video/r2v_extend_video3.mp4"
        ],
        "callback_url": ""
    }
}

Field	Type	Required	Description
prompt	string	Yes	Text prompt describing how to use the reference videos (e.g. `[video 1]`, `[video 2]`, `[video 3]` in order)
model	string	Yes	Model name (e.g. `doubao-seedance-2-0-fast-260128`)
duration	integer	No	Default `5` (2.0) or `-1` (2.5). Allowed: integers from `4` to `15` (2.0) or `4` to `30` (2.5), inclusive, or `-1` for automatic duration selection
size	string	No	Default `720p`. Allowed: `480p`, `720p`; the non-fast 2.0 models also accept `1080p` and `4k`
metadata	object	No	Must include `referenceVideos` for this mode
metadata.ratio	string	No	Default `adaptive`. Allowed: `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, `adaptive`. SeeDance 2.5 accepts `adaptive` only when the reference videos are edited or extended rather than used as style/content references
metadata.referenceVideos	array of string	Yes (this mode)	Up to 3 video URLs on 2.0, up to 10 on 2.5
metadata.callback_url	string	No	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

Constraints on each reference video (and inputs overall):

Count: at most 3 URLs in referenceVideos on 2.0 models, at most 10 on 2.5 models.
Container format: mp4 or mov.
Resolution: 480p or 720p on 2.0 models; 480p, 720p, 1080p or 4k on 2.5 models.
Duration per clip: on 2.0 models each video must be between 2 and 15 seconds (inclusive) and the combined duration of all input videos must not exceed 15 seconds; on 2.5 models each video must be between 2 and 30 seconds and the combined duration must not exceed 30 seconds.
Aspect ratio (width ÷ height): between 0.4 and 2.5 (inclusive).
Width and height (pixels): each between 300 and 6000 (inclusive).
Total pixels (width × height): between 409600 and 927408 (inclusive) on 2.0 models; between 409600 and 8295044 (inclusive) on 2.5 models.
File size: each file must not exceed 50 MB on 2.0 models, or 200 MB on 2.5 models.
Frame rate: between 24 and 60 fps (inclusive).

5. Multimodal References

Combine reference images (images + metadata.imageMode: reference), reference videos (metadata.referenceVideos), and reference audio (metadata.referenceAudio) in one request. Prompts can reference assets by role (e.g. [video 1], [audio 1], [image 1], [image 2]) as in the single-mode sections. Reference videos must still meet the constraints in Reference Videos.

Request body:

{
    "prompt": "First-person framing from [video 1] throughout; use [audio 1] as background music throughout. First-person fruit tea ad for Seedance Ping Ping An An limited-edition apple fruit tea. First frame is [image 1]: your hand picks a dewy Aksu red apple, crisp apple impact sounds. 2-4s: quick cuts — your hands drop apple pieces into a shaker with ice and tea base, shake hard; ice and shaker sounds sync to light drum hits; voiceover: Fresh-cut, shaken fresh. 4-6s: first-person product close-up — layered fruit tea poured into a clear cup, you pipe cream on top and apply a pink label wrap; camera pushes in on cream and tea layers. 6-8s: first-person — you raise the drink from [image 2] toward the camera (as if offering it to the viewer); label clearly visible; voiceover: Take a fresh sip; end frame freezes on [image 2]. Female voice throughout.",
    "model": "doubao-seedance-2-0-fast-260128",
    "images": [
        "https://ark-project.tos-cn-beijing.volces.com/doc_image/r2v_tea_pic1.jpg",
        "https://ark-project.tos-cn-beijing.volces.com/doc_image/r2v_tea_pic2.jpg"
    ],
    "size": "720p",
    "duration": 11,
    "metadata": {
        "imageMode": "reference",
        "referenceVideos": [
            "https://ark-project.tos-cn-beijing.volces.com/doc_video/r2v_tea_video1.mp4"
        ],
        "referenceAudio": [
            "https://ark-project.tos-cn-beijing.volces.com/doc_audio/r2v_tea_audio1.mp3"
        ],
        "ratio": "16:9",
        "callback_url": ""
    }
}

Field	Type	Required	Description
prompt	string	Yes	Long-form prompt referencing `[video N]`, `[audio N]`, `[image N]` as needed
model	string	Yes	Model name (e.g. `doubao-seedance-2-0-fast-260128`)
images	array of string	No	Reference image URLs when `metadata.imageMode` is `reference` (up to 9 on 2.0, 30 on 2.5; see Reference Images)
duration	integer	No	Default `5` (2.0) or `-1` (2.5). Allowed: integers from `4` to `15` (2.0) or `4` to `30` (2.5), inclusive, or `-1` for automatic duration selection
size	string	No	Default `720p`. Allowed: `480p`, `720p`; the non-fast 2.0 models also accept `1080p` and `4k`
metadata	object	No	Combines modes below
metadata.imageMode	string	When using `images` as references	Set to `reference`
metadata.referenceVideos	array of string	No	Reference video URLs (up to 3 on 2.0, 10 on 2.5; constraints in Reference Videos)
metadata.referenceAudio	array of string	No	Reference audio URLs (up to 3 on 2.0, 10 on 2.5). On 2.0 models audio cannot be used alone — at least one reference image or video is required
metadata.ratio	string	No	Default `adaptive`. Allowed: `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, `adaptive`
metadata.callback_url	string	No	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

Constraints on each reference audio clip:

Count: at most 3 URLs in referenceAudio on 2.0 models, at most 10 on 2.5 models.
Format: wav or mp3.
Duration per clip: on 2.0 models each clip must be between 2 and 15 seconds (inclusive) with a combined duration of at most 15 seconds; on 2.5 models each clip must be between 2 and 30 seconds with a combined duration of at most 30 seconds.
File size: each file must not exceed 15 MB, and the whole request body must not exceed 64 MB. Prefer URLs over base64 for large files.
Companion input: on 2.0 models referenceAudio requires at least one reference image or reference video; requests without one are rejected with 400 invalid_request. 2.5 models accept audio on its own.

Generated audio is always mono, regardless of how many channels the input audio has.

6. Audio-Only Reference (SeeDance 2.5)

SeeDance 2.5 is the first family that accepts reference audio with no accompanying image or video. Pass the clip(s) in metadata.referenceAudio and describe the footage that should be generated around them; the model synchronises the visuals to the supplied audio. The same audio constraints as in Multimodal References apply.

Sending this request to a 2.0 model returns 400 invalid_request.

Request body:

{
    "prompt": "A lone busker plays on a rain-slick city street at night, neon signs reflected in the puddles. Use [audio 1] as the performance audio and cut the camera moves to its rhythm; slow push-in on the guitar, then a wide shot as passers-by stop to listen.",
    "model": "doubao-seedance-2-5-260628",
    "duration": 12,
    "size": "720p",
    "metadata": {
        "referenceAudio": [
            "https://example.com/audio/street-guitar.mp3"
        ],
        "ratio": "16:9",
        "callback_url": ""
    }
}

Field	Type	Required	Description
prompt	string	No	Text prompt describing the visuals; refer to the clips as `[audio 1]`, `[audio 2]`, … in order
model	string	Yes	A SeeDance 2.5 model name (e.g. `doubao-seedance-2-5-260628`)
duration	integer	No	Default `-1`. Allowed: integers from `4` to `30`, inclusive, or `-1` for automatic duration selection
size	string	No	Default `720p`. Allowed: `480p`, `720p`
metadata	object	Yes	Must include `referenceAudio` for this mode
metadata.referenceAudio	array of string	Yes (this mode)	Up to 10 audio URLs, each `wav` or `mp3`, 2–30 seconds, 30 seconds combined
metadata.ratio	string	No	Default `adaptive`. Allowed: `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, `adaptive`
metadata.callback_url	string	No	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

7. Video Editing and Extension (SeeDance 2.5)

SeeDance 2.5 can also edit a source clip (restyle it, change its setting, replace its audio while keeping the original motion) or extend it forwards or backwards. Both flows pass the source clip in metadata.referenceVideos and describe the intended change in the prompt.

The task type is inferred from the input assets and the prompt, so write the prompt explicitly — say "keep the original footage and only change …" for an edit, or "continue from the final frame …" for an extension. Both flows have stricter parameter constraints than plain reference-to-video:

metadata.ratio must be adaptive: the output always keeps the aspect ratio of the input video.
For editing, duration must be -1; the output length follows the input (and may be about 0.4 seconds shorter). The input clip must be 4–30 seconds long.
For extending, duration may be set to any value from 4 to 30, or -1.
metadata.output_format: mov is recommended for both the input and the output when the result feeds a professional post-production workflow.

Request body (editing):

{
    "prompt": "Keep the original footage, framing and camera movement exactly as they are; restyle the whole shot as a hand-painted watercolour animation and change the overcast daytime sky to a violet sunset.",
    "model": "doubao-seedance-2-5-260628",
    "duration": -1,
    "metadata": {
        "referenceVideos": [
            "https://example.com/video/street_walk.mov"
        ],
        "ratio": "adaptive",
        "output_format": "mov",
        "callback_url": ""
    }
}

Request body (extending):

{
    "prompt": "Continue [video 1] from its final frame without any cut: the camera keeps pushing down the corridor, the door at the far end swings open and reveals a sunlit courtyard.",
    "model": "doubao-seedance-2-5-260628",
    "duration": 10,
    "metadata": {
        "referenceVideos": [
            "https://example.com/video/corridor.mp4"
        ],
        "ratio": "adaptive",
        "callback_url": ""
    }
}

Field	Type	Required	Description
prompt	string	Yes	Text prompt describing the edit or the continuation
model	string	Yes	A SeeDance 2.5 model name (e.g. `doubao-seedance-2-5-260628`)
duration	integer	No	Editing: must be `-1`. Extension: integers from `4` to `30`, inclusive, or `-1`
size	string	No	Default `720p`. Allowed: `480p`, `720p`
metadata	object	Yes	Must include `referenceVideos` for this mode
metadata.referenceVideos	array of string	Yes (this mode)	Source video URLs; constraints in Reference Videos. Editing requires a clip of 4–30 seconds
metadata.ratio	string	No	Must be `adaptive` (the default) in these modes
metadata.output_format	string	No	`mp4` (default) or `mov`. `mov` is recommended for editing and extension
metadata.callback_url	string	No	If set, attempts to send callbacks to the specified URL. See Callback documentation for more details

1. Submit Video Generation Task

Use the JSON bodies in Usage Examples above; adjust fields for your scenario.

Complete Request:

curl -X POST "https://pay.hkit.ai/v1/video/generations" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer API_KEY" \
  -d @request-body.json

Endpoint:

POST /v1/video/generations

Request Headers:

Parameter	Type	Required	Description
Content-Type	string	Yes	application/json
Authorization	string	Yes	Bearer API_KEY

Response Example:

{
  "task_id": "TASK_ID"
}

Response Field Descriptions:

Field	Type	Description
task_id	string	Task ID for subsequent task status queries

2. Query Task Status

Complete Request:

curl -X GET "https://pay.hkit.ai/v1/video/generations/TASK_ID" \
  -H "Authorization: Bearer API_KEY"

Endpoint:

GET /v1/video/generations/{task_id}

Request Headers:

Parameter	Type	Required	Description
Authorization	string	Yes	Bearer API_KEY

Path Parameters:

Parameter	Type	Required	Description
task_id	string	Yes	Task ID

Response Example (Processing):

{
    "code": "success",
    "message": "",
    "data": {
        "task_id": "<TASK_ID>",
        "action": "generate",
        "status": "IN_PROGRESS",
        "fail_reason": "",
        "submit_time": 1776717808,
        "start_time": 1776717809,
        "finish_time": 0,
        "progress": "30%",
        "data": {
            "created_at": 1776717808,
            "draft": false,
            "execution_expires_after": 172800,
            "generate_audio": true,
            "id": "<TASK_ID>",
            "model": "doubao-seedance-2-0-fast-260128",
            "service_tier": "default",
            "status": "running",
            "updated_at": 1776717808
        }
    }
}

Response Example (Success):

Note: When the task succeeds, the data.fail_reason field will contain the video download URL instead of an error message — this is a known behavior of the upstream SeeDance API. The recommended way to retrieve the video URL is via data.data.content.video_url.

{
    "code": "success",
    "message": "",
    "data": {
        "task_id": "<TASK_ID>",
        "action": "generate",
        "status": "SUCCESS",
        "fail_reason": "<VIDEO_URL>",
        "submit_time": 1776717808,
        "start_time": 1776717818,
        "finish_time": 1776717962,
        "progress": "100%",
        "data": {
            "content": {
                "video_url": "<VIDEO_URL>"
            },
            "created_at": 1776717808,
            "draft": false,
            "duration": 4,
            "execution_expires_after": 172800,
            "framespersecond": 24,
            "generate_audio": true,
            "id": "<TASK_ID>",
            "model": "doubao-seedance-2-0-fast-260128",
            "ratio": "16:9",
            "resolution": "480p",
            "seed": 93719,
            "service_tier": "default",
            "status": "succeeded",
            "updated_at": 1776717961,
            "usage": {
                "completion_tokens": 40594,
                "total_tokens": 40594
            }
        }
    }
}

Note: The returned duration is an integer approximation, calculated as the total frame count divided by 24 and rounded down. A 133-frame video lasts 5.54 seconds but reports a duration of 5.

3. Callback

POST <YOUR_CALLBACK_URL>

The webhook requests consist of "wrapper" data, taking the following format:

{
  "webhook_type": "VIDEO_TASK",
  "payload": {
    "channel_type": "DoubaoVideo",
    "data": {}
  }
}

The payload.data field contains the data in the same format that the 2. Query Task Status endpoint's response.data would provide.

Request Example (Task Running)

{
  "webhook_type": "VIDEO_TASK",
  "payload": {
    "channel_type": "DoubaoVideo",
    "data": {
      "task_id": "<TASK_ID>",
      "action": "generate",
      "status": "IN_PROGRESS",
      "fail_reason": "",
      "submit_time": 1776717808,
      "start_time": 1776717809,
      "finish_time": 0,
      "progress": "30%",
      "data": {
        "created_at": 1776717808,
        "draft": false,
        "execution_expires_after": 172800,
        "generate_audio": true,
        "id": "<TASK_ID>",
        "model": "doubao-seedance-2-0-fast-260128",
        "service_tier": "default",
        "status": "running",
        "updated_at": 1776717808
      }
    }
  }
}

Request Example (Task Completed Successfully)

{
  "webhook_type": "VIDEO_TASK",
  "payload": {
    "channel_type": "DoubaoVideo",
    "data": {
      "task_id": "<TASK_ID>",
      "action": "generate",
      "status": "SUCCESS",
      "fail_reason": "<VIDEO_URL>",
      "submit_time": 1776717808,
      "start_time": 1776717818,
      "finish_time": 1776717962,
      "progress": "100%",
      "data": {
        "content": {
          "video_url": "<VIDEO_URL>"
        },
        "created_at": 1776717808,
        "draft": false,
        "duration": 4,
        "execution_expires_after": 172800,
        "framespersecond": 24,
        "generate_audio": true,
        "id": "<TASK_ID>",
        "model": "doubao-seedance-2-0-fast-260128",
        "ratio": "16:9",
        "resolution": "480p",
        "seed": 93719,
        "service_tier": "default",
        "status": "succeeded",
        "updated_at": 1776717961,
        "usage": {
          "completion_tokens": 40594,
          "total_tokens": 40594
        }
      }
    }
  }
}

SeeDance Video API Documentation

On this page