Seedance 2.0

POST/api/v1/jobs/createTask

Seedance 2.0 is ByteDance's video generation model. It supports text-to-video, first/last-frame image-to-video, and multimodal references, produces 4 to 15 seconds of 480p or 720p video, and can generate a matching audio track.

Set model to seedance-2.0. The request is asynchronous: store the returned taskId, then query the task or wait for the callback.

Request example

cURL

curl -X POST https://api.xbaiapi.com/api/v1/jobs/createTask \
  -H "Authorization: Bearer xb-xxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.0",
    "input": {
      "prompt": "a cat stretching on a sunlit windowsill",
      "resolution": "720p",
      "duration": 5,
      "ratio": "16:9",
      "generateAudio": true,
      "imageUrls": []
    },
    "callBackUrl": "https://your-server.com/webhook"
  }'

Parameters

Field	Type	Required	Notes
`model`	string	Yes	Fixed value `seedance-2.0`
`input.prompt`	string	Conditional	Up to 20480 characters; optional for first/last-frame image-to-video
`input.resolution`	string	No	`480p` or `720p`, defaults to `720p`
`input.duration`	number	No	Integer seconds from 4 to 15, defaults to 5. Automatic duration `-1` is currently unavailable
`input.ratio`	string	No	`adaptive`, `16:9`, `9:16`, `1:1`, `4:3`, `3:4`, or `21:9`. Defaults to `adaptive`
`input.generateAudio`	boolean	No	Defaults to `true`
`input.firstFrameUrl`	string	No	First frame; providing it selects image-to-video and is required in that mode
`input.lastFrameUrl`	string	No	Optional last frame; requires a first frame
`input.imageUrls`	string[]	No	Up to 9 multimodal reference images, up to 30 MB each
`input.videoUrls`	string[]	No	Up to 3 reference videos, 2 to 15 seconds and up to 50 MB each; total duration at most 15 seconds
`input.audioUrls`	string[]	No	Up to 3 reference audio clips, 2 to 15 seconds and up to 50 MB each; total duration at most 15 seconds; requires a reference image or video
`input.webSearch`	boolean	No	Text-to-video only; defaults to `false`
`input.realPersonMode`	boolean	No	Image-to-video and multimodal only; defaults to `true`
`input.conversionSlots`	string[]	No	Asset conversion slots; defaults to `["all"]`; image-to-video accepts `firstFrameUrl` and `lastFrameUrl`; multimodal accepts `image1` through `image9` and `video1` through `video3`
`input.returnLastFrame`	boolean	No	Request a last frame; defaults to `false`. The public task result still returns only the video
`input.seed`	number	No	Integer from `-1` to `2147483647`; defaults to `-1`
`callBackUrl`	string	No	Omit it and poll the query endpoint instead

Without media, the request uses text-to-video. firstFrameUrl selects first/last-frame image-to-video. Any reference media list selects multimodal generation. First/last frames cannot be combined with reference lists. Media can be a publicly reachable HTTP(S) URL or a Base64 media data URI. File sizes and durations are checked by the generation service.

Your account is charged by output duration. See pricing for CNY Token reference prices.

Response

Success response

{
  "code": 200,
  "msg": "success",
  "data": {
    "taskId": "xb_a1b2c3d4e5f6...",
    "creditsCharged": 165
  }
}

When the task succeeds, read the video URL from resultJson.resultUrls in the query response.

Request example

Parameters

Response

On this page