Seedance Video API

Use this page when you want to create or query Seedance video generation tasks through Exchange Token.

Supported models and endpoints

Item	Value
Public models	doubao-seedance-2-0-260128, seedance-2.0, seedance-2.0-fast
Create task	POST /api/v3/contents/generations/tasks
Query task	GET /api/v3/contents/generations/tasks/{id}
Base URL example	https://api.exchangetoken.ai
Recommended auth	Authorization: Bearer <YOUR_API_KEY>

Request body

Field	Type	Required	Description
model	string	Yes	Exact model name enabled for your account, such as doubao-seedance-2-0-260128 or seedance-2.0
content	object[]	Yes	Input array
generate_audio	boolean	No	Default true
tools	object[]	No	Environment-dependent capability
resolution	string	No	480p or 720p
ratio	string	No	21:9, 16:9, 4:3, 1:1, 3:4, 9:16, adaptive
duration	integer	No	4 to 15, or -1
watermark	boolean	No	Follow upstream capability

Avoid depending on service_tier, draft, frames, or camera_fixed.

Model selection

Use doubao-seedance-2-0-260128 when you need compatibility with the domestic Doubao/Seedance 2.0 model ID. Use seedance-2.0 when your account is configured for the Exchange Token generic Seedance alias. Fast-tier access through seedance-2.0-fast is account-dependent.

Domestic-compatible and international-compatible Seedance pools should use different public model names. Do not assume assets or tasks created for one compatibility pool are valid for another unless your account manager has enabled that workflow.

Asset-backed domestic 2.0 requests

To use your own image, video, or audio media with doubao-seedance-2-0-260128, create an Exchange Token Seedance asset first, then reference it in the video task as asset://<ASSET_ID>.

CreateAssetGroup and CreateAsset are Exchange Token Seedance asset gateway endpoints. They are not OpenAI file upload endpoints and they do not accept multipart file uploads. CreateAsset currently takes a public HTTPS media URL and returns the stable asset ID that video requests should reference.

Minimal flow:

1. Call POST /open/CreateAssetGroup and save GROUP_ID
2. Call POST /open/CreateAsset with GroupId and a public media URL, then save ASSET_ID
3. Poll POST /open/GetAsset until Result.Status = "Active"
4. Submit the video task with asset://$ASSET_ID

Example video request:

curl -sS "$ET_BASE/api/v3/contents/generations/tasks" \
  -H "Authorization: Bearer $ET_TOKEN" \
  -H "Content-Type: application/json" \
  -d "{
    \"model\":\"doubao-seedance-2-0-260128\",
    \"content\":[
      {\"type\":\"text\",\"text\":\"Make the character smile naturally and gently wave while the camera slowly pushes in\"},
      {\"type\":\"image_url\",\"image_url\":{\"url\":\"asset://$ASSET_ID\"},\"role\":\"first_frame\"}
    ],
    \"duration\":5,
    \"watermark\":false
  }"

Supported content and roles

Content types

type	Description
text	Prompt text
image_url	Image input
video_url	Video input
audio_url	Audio input

Roles

Input type	Role	Description
image_url	first_frame	First-frame image-to-video
image_url	last_frame	Last-frame image-to-video
image_url	reference_image	Multimodal reference image
video_url	reference_video	Reference video
audio_url	reference_audio	Reference audio

Supported URL forms

Input type	URL forms
image_url	Public URL, Base64, asset://<ASSET_ID>
video_url	Public URL, asset://<ASSET_ID>
audio_url	Public URL, Base64, asset://<ASSET_ID>

If you plan to reference asset://<ASSET_ID>, create assets first on the Seedance Asset API page.

Valid request patterns

Scenario	Minimum input
Text to video	text
First-frame image to video	image_url(first_frame) with optional text
First and last frame image to video	image_url(first_frame) + image_url(last_frame) with optional text
Multimodal guided generation	Legal combinations of reference_image, reference_video, reference_audio, with optional text
Video editing	reference_video with optional reference_image and text
Video extension	reference_video with optional text

Request rules and media limits

Rules

• first_frame and last_frame mode cannot be mixed with reference_image mode
• Audio cannot be submitted alone
• asset://<ASSET_ID> is only valid after the asset status becomes Active
• tools=[{"type":"web_search"}] should be validated before production use

Media limits

Input	Limits
Image	jpeg, png, webp, bmp, tiff, gif; aspect ratio (0.4, 2.5); 300 to 6000 px; < 30 MB each; request body <= 64 MB
Video	mp4, mov; 480p or 720p; 2 to 15 s each; max 3 clips; total duration <= 15 s; 24 to 60 FPS; <= 50 MB each
Audio	wav, mp3; 2 to 15 s each; max 3 clips; total duration <= 15 s; <= 15 MB each; request body <= 64 MB

Task responses

Create response

{
  "id": "sg_vid_m9xk3g_0f8a4c0d3a2b4e9f1c20"
}

Exchange Token returns its own queryable task ID. Use it for later polling.

Query response

{
  "id": "sg_vid_m9xk3g_0f8a4c0d3a2b4e9f1c20",
  "model": "doubao-seedance-2-0-260128",
  "status": "succeeded",
  "content": {
    "video_url": "https://example.com/output.mp4?sig=foo&x=1"
  },
  "usage": {
    "completion_tokens": 108900,
    "total_tokens": 108900
  },
  "created_at": 1776209922,
  "updated_at": 1776210072,
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "generate_audio": true
}

Typical statuses:

• queued
• running
• succeeded
• failed
• cancelled

The response model may show the submitted model name or an upstream versioned model ID.

Error format

{
  "error": {
    "type": "invalid_request",
    "message": "..."
  }
}

HTTP status	error.type	Typical meaning
400	invalid_request	Invalid parameters or JSON
401	unauthorized	Missing or invalid API key
402	quota_exhausted	Video quota exhausted
403	forbidden	Token disabled, expired, or not allowed
404	task_not_found	Task not found
429	rate_limited	Too many requests
502	upstream_error	Upstream submit or parse failure
503	no_channel / async_task_unavailable	No available channel or async task unavailable

Examples

Before running the examples below:

export ET_BASE='https://api.exchangetoken.ai'
export ET_TOKEN='sk-xxxxxx'
export ASSET_ID='asset-xxxxxxxx'
export TASK_ID='sg_vid_xxxxxxxx'

Text to video

curl -sS "$ET_BASE/api/v3/contents/generations/tasks" \
  -H "Authorization: Bearer $ET_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model":"doubao-seedance-2-0-260128",
    "content":[
      {"type":"text","text":"A short-haired woman turns back and smiles on a city street, cinematic style, stable forward camera movement"}
    ],
    "duration":5,
    "resolution":"720p",
    "watermark":false
  }'

Asset-backed first-frame generation

curl -sS "$ET_BASE/api/v3/contents/generations/tasks" \
  -H "Authorization: Bearer $ET_TOKEN" \
  -H "Content-Type: application/json" \
  -d "{
    \"model\":\"doubao-seedance-2-0-260128\",
    \"content\":[
      {\"type\":\"text\",\"text\":\"Make the character nod slightly and wave to the camera\"},
      {\"type\":\"image_url\",\"image_url\":{\"url\":\"asset://$ASSET_ID\"},\"role\":\"first_frame\"}
    ],
    \"duration\":5,
    \"watermark\":false
  }"

Query a task

curl -sS "$ET_BASE/api/v3/contents/generations/tasks/$TASK_ID" \
  -H "Authorization: Bearer $ET_TOKEN"

Video extension

curl -sS "$ET_BASE/api/v3/contents/generations/tasks" \
  -H "Authorization: Bearer $ET_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
      "model": "seedance-2.0-fast",
      "content": [
        {
          "type": "text",
          "text": "Generate the content after [Video1]: the two men who are late run towards them, the five people finally meet and have a friendly chat."
        },
        {
          "type": "video_url",
          "video_url": {
            "url": "https://ark-doc.tos-ap-southeast-1.bytepluses.com/doc_image/video_edit_prolong_ref_video.mp4"
          },
          "role": "reference_video"
        }
      ],
      "generate_audio": true,
      "ratio": "16:9",
      "duration": 5,
      "watermark": true
    }'

Related page

• Seedance Asset API