Seedance 2.5 Video Generation API

Generate video with or without audio from text, images, video, and audio. Async task API: submit a request, receive a task ID, then poll for status to get the result.

Generate 1 target video with Doubao Seedance 2.5, with or without audio. Up to 30 seconds, 480p / 720p. Supported scenarios:

Text-to-video: generate 1 target video from a text prompt
Image-to-video (first frame): 1 first-frame image + optional text prompt
Image-to-video (first & last frames): first-frame image + last-frame image + optional text prompt
Omni-reference-to-video: reference images (0–30) + reference videos (0–10) + reference audio (0–10) + optional text prompt. Audio-only is allowed. Supports creating a new video, editing, and extending, including 30-second one-shot output

This is an async task-based API: submit a request to receive a task id, poll for status, then download the video when complete.

Only the Doubao Seedance 2.5 model (doubao-seedance-2-5-260628) is supported.

Note: First frame, first & last frames, and omni-reference-to-video are 3 mutually exclusive scenarios and must not be mixed. Omni-reference can treat a reference image as the first / last frame via the prompt; for strict first/last-frame matching, prefer image-to-video (first & last frames).

Base URL

https://api.autorouter.top — Production

Authentication

BearerAuth: http (bearer) Authenticate using a Bearer Token. Format: Authorization: Bearer sk-xxxxxx

Endpoints

POST `/api/v3/contents/generations/tasks`

Create Video Generation Task

Create a video generation task from text, images, video, and audio.

Request Body

Content-Type: application/json

Example (text-to-video):

{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "A kitten yawns at the camera"
    }
  ],
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "generate_audio": true,
  "watermark": false
}

Example (image-to-video, first frame):

{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Make this cat jump"
    },
    {
      "type": "image_url",
      "image_url": {
        "url": "https://example.com/cat.jpg"
      },
      "role": "first_frame"
    }
  ],
  "resolution": "720p",
  "ratio": "adaptive",
  "duration": 5,
  "generate_audio": true
}

Example (omni-reference-to-video):

{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Extend this video so the cat walks out of frame"
    },
    {
      "type": "video_url",
      "video_url": {
        "url": "https://example.com/cat-original.mp4"
      },
      "role": "reference_video"
    }
  ],
  "resolution": "720p",
  "ratio": "adaptive",
  "duration": -1,
  "omni_reference_task_type": "extend",
  "generate_audio": true
}

Properties:

Name	Type	Required	Description
`model`	string	Yes	Model ID. Fixed to `doubao-seedance-2-5-260628`
`content`	object[]	Yes	Input content list: text, image, video, audio
`content[].type`	string	Yes	Content type: `text` / `image_url` / `video_url` / `audio_url`
`content[].text`	string	Yes*	Text prompt (required when `type=text`)
`content[].image_url.url`	string	Yes*	Image URL / Base64 / asset ID (when `type=image_url`)
`content[].video_url.url`	string	Yes*	Video URL / asset ID (when `type=video_url`)
`content[].audio_url.url`	string	Yes*	Audio URL / Base64 / asset ID (when `type=audio_url`)
`content[].role`	string	Conditional	Media role; see below
`resolution`	string	No	Video resolution. Default `720p`; `480p` or `720p`
`ratio`	string	No	Aspect ratio. Default `adaptive`. See below
`duration`	integer	No	Duration in seconds. Range `[4, 30]` or `-1`; default `-1`
`generate_audio`	boolean	No	Whether to generate audio. `true` (default) with audio; `false` silent
`omni_reference_task_type`	string	No	Omni-reference task-type hint. Default `auto`
`output_format`	string	No	Output format. Default `mp4`; `mp4` or `mov`
`watermark`	boolean	No	Whether to add an "AI generated" watermark. `false` (default) off; `true` on
`callback_url`	string	No	Callback URL for task status changes
`return_last_frame`	boolean	No	Whether to return the last-frame image. Default `false`
`execution_expires_after`	integer	No	Task timeout in seconds. Range `[3600, 259200]`; default `172800` (48 hours)
`priority`	integer	No	Queue priority. Range `[0, 9]`; default `0`. Higher values run first
`tools`	object[]	No	Tool config such as web search. See below
`safety_identifier`	string	No	End-user identifier (ASCII, ≤64 chars) for safety auditing

Note: besides the request body, resolution, ratio, duration, and watermark can also be appended to the prompt as weakly validated --[parameters] (e.g. --rs 720p --rt 16:9 --dur 5 --wm false). Prefer request-body strong validation.

Seedance 2.5 does not accept reference images / videos that contain real human faces. Use platform virtual portraits, licensed real-person assets, or original face-containing outputs generated by Seedance 2.5 / 2.0 on this account within the last 30 days.

modelstring(Required)

Model ID to call.

Model	Description	Default resolution	Supported resolutions
`doubao-seedance-2-5-260628`	Seedance 2.5; text / image / omni-reference; up to 30 seconds	`720p`	`480p`, `720p`

contentobject[](Required)

Input used to generate the video: text, images, audio, and video.

Supported combinations (text is optional except for text-to-video): text only; text + image; text + video; text + audio (audio-only is allowed); text + image + audio; text + image + video; text + video + audio; text + image + video + audio.

typestring(Required)

Content type: text, image_url, video_url, audio_url.

textstring(Conditional)

Text prompt describing the desired video. Required for text-to-video; optional for image-to-video and omni-reference.

Languages: Chinese and English; also Spanish, Indonesian, Portuguese, Japanese, Malay, Thai, Arabic, Vietnamese, Korean
Length: up to 500 Chinese characters or 1000 English words. Too much text can dilute the intent
For videos with audio, put dialogue in double quotes, e.g. A man stops a woman and says: "Remember, never point at the moon with your finger."

Example: A kitten yawns at the camera

image_url.urlstring(Conditional)

Image source. Supports:

Public URL: a publicly reachable image URL
Base64: data:image/<format>;base64,<encoded>; format must be lowercase, e.g. data:image/png;base64,...
Asset ID: asset://<ASSET_ID> (preset assets / virtual portraits)

Per-image limits:

Formats: jpeg, png, webp, bmp, tiff, gif, heic, heif
Aspect ratio (w/h): [0.4, 2.5]
Side length: [300, 6000] px
Size: < 30 MB each; request body ≤ 64 MB. Avoid Base64 for large files

Count:

Image-to-video (first frame): 1 image
Image-to-video (first & last frames): 2 images
Omni-reference-to-video: 1–30 images

video_url.urlstring(Conditional)

Reference video source: public URL or asset ID (asset://<ASSET_ID>).

Per-video limits:

Formats: mp4, mov (video H.264/AVC, H.265/HEVC; audio AAC, MP3)
Resolution: 480p, 720p, 1080p, 4k
Duration: [2, 30] s each; up to 10 reference videos; total duration of all videos ≤ 30 s
Aspect ratio: [0.4, 2.5]; side length [300, 6000] px
Total pixels: [409600, 8295044]
Size: ≤ 200 MB each
Frame rate: [24, 60] FPS

audio_url.urlstring(Conditional)

Reference audio source: public URL, Base64 (data:audio/<format>;base64,...), or asset ID (asset://<ASSET_ID>).

Per-audio limits:

Formats: wav, mp3
Duration: [2, 30] s each; up to 10 clips; total duration of all audio ≤ 30 s
Size: ≤ 15 MB each; request body ≤ 64 MB. Avoid Base64 for large files

Seedance 2.5 can take audio alone, without images / video, or together with images / video.

rolestring(Conditional)

Media position or purpose.

Image-to-video (first frame)

One image_url object; role is first_frame or omitted.

Image-to-video (first & last frames)

Two image_url objects; role is required:

First frame: first_frame
Last frame: last_frame

Note: first and last frame images may be the same. If aspect ratios differ, the first frame wins and the last frame is cropped to match.

Omni-reference-to-video

Reference image: role required, always reference_image
Reference video: role always reference_video
Reference audio: role always reference_audio

Note: first frame, first & last frames, and omni-reference-to-video are 3 mutually exclusive scenarios and must not be mixed.

resolutionstring(Optional)

Video resolution. Default 720p; 480p or 720p.

Seedance 2.5 does not support 1080p or 4k.

ratiostring(Optional)

Aspect ratio of the generated video. Default adaptive.

Allowed values: 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive

Note: for Seedance 2.5 video edit, video extend, and first-frame / first-and-last-frame tasks (see task types and matching rules), ratio must be adaptive; a specific aspect ratio is not allowed.

Limits by task type:

Task type	Rule
Text-to-video	You may set an aspect ratio, or let the model choose from the prompt (`adaptive` allowed)
Image-to-video (first / first & last frames)	Output matches the first-frame image aspect ratio; `adaptive` only
Omni-reference-to-video	You may set an aspect ratio, or let the model choose (`adaptive` allowed)
Omni edit / extend	Output matches the source video aspect ratio; `adaptive` only

Pixel sizes by resolution:

Resolution	16:9	4:3	1:1	3:4	9:16	21:9
480p	854×480	752×560	640×640	560×752	480×854	992×432
720p	1280×720	1112×834	960×960	834×1112	720×1280	1470×630

If the chosen aspect ratio does not match the uploaded image, the platform center-crops.

durationinteger(Optional)

Output duration in seconds. Default -1; range [4, 30] or -1.

When duration = -1, the model picks an integer second length within the valid range.

Video-edit limits:

duration must be -1; a specific output length is not allowed
The source video to edit must be [4, 30] s, or the request fails
Output duration is essentially the same as the input (may be slightly shorter, ~0.4 s)

The duration returned by the query API is an integer-second estimate (floored): returned duration = actual frame count / 24.

generate_audioboolean(Optional)

Whether the output video includes sound synchronized with the picture. Default true.

true: video with audio; the model generates matching speech, SFX, and BGM from the prompt and visuals
false: silent video

Note: generated audio is always mono, regardless of the input audio channel count.

omni_reference_task_typestring(Optional)

Task-type hint for omni-reference-to-video, used to validate constraints early. Default auto.

auto: the model infers the task type from assets and the prompt. If parameters are incompatible with the actual type, the task fails asynchronously (InvalidParameter.TaskTypeConstraint)
reference: reference-to-video; generate a new video from reference images, videos, or audio. No extra ratio / duration limits
edit: video edit; edit the picture or audio of the source video. content must include at least one reference_video of 4–30 seconds; ratio must be adaptive; duration must be -1
extend: video extend; extend the source video forward or backward. content must include at least one reference_video; ratio must be adaptive

Note: at runtime the model still infers the task type from the prompt. If that differs from the value you set, the task still fails asynchronously (InvalidParameter.TaskTypeMismatch). Follow the prompt style for each task type to reduce errors.

This parameter applies only to omni-reference-to-video.

output_formatstring(Optional)

Output video format. Default mp4.

mp4: general-purpose, best compatibility, standard color precision; plays on web, mobile, players, and distribution platforms
mov: high color-precision format for professional work (H.264 + yuv444p + PCM); better color and brightness consistency for grading, keying, and compositing. Prefer mov as input and output for edit and extend

Note: some players may not support mov. Common players that do: IINA (macOS), VLC, mpv, ffplay.

toolsobject[](Optional)

Tools the model may call.

typestring(Required)

Tool type. Currently web_search. When enabled, the model decides whether to search the web from the prompt, which can improve freshness at some extra latency.

Actual search count is in usage.tool_usage.web_search from the query API; 0 means no search.

Example:

{
  "tools": [
    { "type": "web_search" }
  ]
}

Responses

200: Task created

Content-Type: application/json

400: Invalid request parameters

Content-Type: application/json

429: Rate limited

Content-Type: application/json

Request examples

curl -X POST "https://api.autorouter.top/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "A kitten yawns at the camera"
    }
  ],
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "generate_audio": true,
  "watermark": false
}'

Response examples

{
  "id": "cgt-2025xxxxxx-xxxxx"
}

Response fields:

Name	Type	Description
`id`	string	Video generation task ID. Kept for 7 days from `created_at`. Creation is async; use the query API to get the result

More Scenario Examples

Image-to-video (first frame)

curl -X POST "https://api.autorouter.top/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Make this cat jump"
    },
    {
      "type": "image_url",
      "image_url": {
        "url": "https://example.com/cat.jpg"
      },
      "role": "first_frame"
    }
  ],
  "resolution": "720p",
  "ratio": "adaptive",
  "duration": 5,
  "generate_audio": true
}'

Image-to-video (first & last frames)

curl -X POST "https://api.autorouter.top/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Slowly pull the camera back from a close-up"
    },
    {
      "type": "image_url",
      "image_url": {
        "url": "https://example.com/first.jpg"
      },
      "role": "first_frame"
    },
    {
      "type": "image_url",
      "image_url": {
        "url": "https://example.com/last.jpg"
      },
      "role": "last_frame"
    }
  ],
  "resolution": "720p",
  "ratio": "adaptive",
  "duration": 5,
  "generate_audio": true
}'

Omni-reference-to-video

curl -X POST "https://api.autorouter.top/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Using these references, generate a video of a cat playing in the living room"
    },
    {
      "type": "image_url",
      "image_url": {
        "url": "https://example.com/ref-cat.jpg"
      },
      "role": "reference_image"
    },
    {
      "type": "video_url",
      "video_url": {
        "url": "https://example.com/ref-motion.mp4"
      },
      "role": "reference_video"
    },
    {
      "type": "audio_url",
      "audio_url": {
        "url": "https://example.com/ref-voice.mp3"
      },
      "role": "reference_audio"
    }
  ],
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "omni_reference_task_type": "reference",
  "generate_audio": true
}'

Video edit

curl -X POST "https://api.autorouter.top/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Replace the background with a seaside sunset"
    },
    {
      "type": "video_url",
      "video_url": {
        "url": "https://example.com/cat-original.mp4"
      },
      "role": "reference_video"
    }
  ],
  "resolution": "720p",
  "ratio": "adaptive",
  "duration": -1,
  "omni_reference_task_type": "edit",
  "generate_audio": true
}'

Video extend

curl -X POST "https://api.autorouter.top/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Extend this video so the cat walks out of frame"
    },
    {
      "type": "video_url",
      "video_url": {
        "url": "https://example.com/cat-original.mp4"
      },
      "role": "reference_video"
    }
  ],
  "resolution": "720p",
  "ratio": "adaptive",
  "duration": -1,
  "omni_reference_task_type": "extend",
  "generate_audio": true
}'

Audio only

curl -X POST "https://api.autorouter.top/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxx" \
  -d '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "Generate matching visuals from this audio"
    },
    {
      "type": "audio_url",
      "audio_url": {
        "url": "https://example.com/narration.mp3"
      },
      "role": "reference_audio"
    }
  ],
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "omni_reference_task_type": "reference",
  "generate_audio": true
}'

GET `/api/v3/contents/generations/tasks/{id}`

Query Video Generation Task

Query the status and result of a video generation task.

Only tasks from the last 7 days are queryable, in [T-7 days, T), where T is the UTC timestamp of the request (second precision)
Video URLs are valid for 24 hours; download or transfer promptly
Seedance 2.5 video URLs may be downloaded at most 100 times

Headers

Name	Type	Required	Description
`Authorization`	string	Yes	Auth header. Format: `Bearer sk-xxxxxx`

Path Parameters

Name	Type	Required	Description
`id`	string	Yes	Video generation task ID to query

Responses

200: Task queried

Content-Type: application/json

Request examples

curl -X GET "https://api.autorouter.top/api/v3/contents/generations/tasks/{id}" \
  -H "Authorization: Bearer sk-xxxxxx"

Response examples

{
  "id": "cgt-2025xxxxxx-xxxxx",
  "model": "doubao-seedance-2-5-260628",
  "status": "succeeded",
  "error": null,
  "created_at": 1730000000,
  "updated_at": 1730000060,
  "content": {
    "video_url": "https://ark-video-xxx.volces.com/xxx.mp4",
    "last_frame_url": "https://ark-video-xxx.volces.com/xxx.png"
  },
  "seed": -1,
  "resolution": "720p",
  "ratio": "16:9",
  "duration": 5,
  "framespersecond": 24,
  "generate_audio": true,
  "output_format": "mp4",
  "tools": [
    { "type": "web_search" }
  ],
  "safety_identifier": "user_hash_xxx",
  "priority": 0,
  "service_tier": "default",
  "execution_expires_after": 172800,
  "usage": {
    "completion_tokens": 108000,
    "total_tokens": 108000,
    "tool_usage": {
      "web_search": 1
    }
  }
}

Response fields:

Name	Type	Description
`id`	string	Video generation task ID
`model`	string	Model name and version used, as `name-version`
`status`	string	Task status; see below
`error`	object / null	Error info; `null` on success, error payload on failure
`created_at`	integer	Task creation time, Unix timestamp (seconds)
`updated_at`	integer	Last status update time, Unix timestamp (seconds)
`content`	object	Output of the video generation task
`content.video_url`	string	Generated video URL. Valid 24 hours; download limit 100
`content.last_frame_url`	string	Last-frame image URL. Valid 24 hours; download limit 100. Returned only when create used `"return_last_frame": true`
`seed`	integer	Seed integer used for this request
`resolution`	string	Output video resolution
`ratio`	string	Output video aspect ratio
`duration`	integer	Output video duration (seconds)
`framespersecond`	integer	Output video frame rate
`generate_audio`	boolean	Whether the output includes synchronized audio
`output_format`	string	Actual output format (`mp4` or `mov`)
`tools`	object[]	Tools actually used. Omitted when none were used
`tools[].type`	string	Tool type used, e.g. `web_search`
`safety_identifier`	string	End-user identifier. Echoed only if set on create
`priority`	integer	Execution priority assigned by account and model policy. Higher runs first
`service_tier`	string	Service tier actually used
`execution_expires_after`	integer	Expiry after submit (seconds). Timed-out tasks become `expired`
`usage`	object	Token usage for this request
`usage.completion_tokens`	integer	Tokens consumed to generate the video; usable for billing reconciliation
`usage.total_tokens`	integer	Total tokens for the request. Video generation does not count input tokens, so `total_tokens` = `completion_tokens`
`usage.tool_usage`	object	Tool usage counters
`usage.tool_usage.web_search`	integer	Web-search calls; returned only when web search was enabled

statusstring

Task status:

queued: queued
running: running
cancelled: cancelled (removed 24h after cancel; only queued tasks can be cancelled)
succeeded: succeeded
failed: failed
expired: timed out

Poll every 3–5 seconds until status is succeeded, failed, cancelled, or expired.

errorobject / null

Error message. null on success; error payload on failure.

Note: for some Seedance 2.5 task types, errors are returned only after the task is dequeued and consumed.

codestring

Error code.

messagestring

Error message.

contentobject

Output of the video generation task.

video_urlstring

Generated video URL. Valid for 24 hours; download limit 100. Download or transfer promptly.

last_frame_urlstring

Last-frame image URL. Valid for 24 hours; download limit 100. Download or transfer promptly.

Returned only when create used "return_last_frame": true.

generate_audioboolean

Whether the output video includes synchronized audio.

true: video includes synchronized audio
false: silent video

output_formatstring

Actual output video format.

mp4: general-purpose format
mov: professional post-production format

usageobject

Token usage for this request.

completion_tokensinteger

Tokens consumed to generate the video; usable for billing reconciliation.

total_tokensinteger

Total tokens for the request. Video models do not count input tokens (0), so total_tokens = completion_tokens.

tool_usageobject

Tool usage counters.

web_search (integer): number of web-search calls; returned only when web search was enabled

Error handling

HTTP 400

Case	Suggestion
Missing `content` / `model`	Provide required fields
Unknown model name	Use `doubao-seedance-2-5-260628`
`1080p` or `4k`	Seedance 2.5 supports only `480p` and `720p`
Mixing first frame / first & last frames / reference-to-video	The three scenarios are mutually exclusive
Non-`adaptive` `ratio` on first frame / first & last frames / edit / extend	Set `ratio` to `adaptive`
Specific `duration` on video edit	Edit tasks only allow `duration = -1`
Too many reference videos / audio clips or total duration too long	Up to 10 videos and 10 audio clips; each total duration ≤ 30 s
`omni_reference_task_type` does not match assets / prompt	Meet constraints such as `reference_video` for that type, or use `auto`

HTTP 401 / 403

401 Unauthorized: Invalid or expired API Key
403 Forbidden: API Key not allowed for this model (check the token model allowlist)

HTTP 402

Insufficient balance. Top up in the AutoRouter console.

Task `failed` / `expired`

Cause	Suggestion
Content moderation	Adjust the prompt or replace reference assets
Media URL unreachable	Ensure the URL is publicly reachable, or use Base64 / asset ID
Unauthorized real human faces	Use virtual portraits, licensed assets, or recent outputs from this account
Task-type constraints not met	Check `ratio`, `duration`, and reference-video duration for that task type
Timeout	Increase `execution_expires_after` and retry

Failed tasks are automatically refunded.

Seedance 2.5 260628 Video Generation

modelstring(Required)

contentobject[](Required)

Properties

resolutionstring(Optional)

ratiostring(Optional)

durationinteger(Optional)

generate_audioboolean(Optional)

omni_reference_task_typestring(Optional)

output_formatstring(Optional)

toolsobject[](Optional)

Properties

statusstring

errorobject / null

Properties

contentobject

Properties

generate_audioboolean

output_formatstring

usageobject

Properties

On this page