seedance-2.5-cl

seedance-2.5-cl is a MoonNexAI Seedance 2.5 model for text-to-video, image-to-video and mixed image, video and audio references. Set model to seedance-2.5-cl.

Current Configuration

Item	Annotations
Create Entry	POST /v1/videos;official shape compatible entrance POST /api/v3/contents/generations/tasks
Query Entry	GET /v1/videos/{task_id} or GET /v1/tasks/{task_id}
Output Format	mp4, mov
resolution	480P, 720P, 1080P
duration	4 - 30 seconds
aspect ratio	16:9, 9:16, 1:1, 4:3, 3:4, 21:9; part task for adaptive
Generate Audio	Support, default generate_audio on

The actual available parameters are based on GET /v1/models and the current API Key permissions.

Capacity boundaries

Capacity	Support
text-to-video	Support
Single and multi-graph reference	Up to 30 pictures
reference video	Maximum 10
reference audio	maximum 10; pictures or videos are required at the same time
Mixed reference media assets	Total top 50
First / last frame	Support, by role in request
Realist reference.	Support; content security clearance still required

Create a video task

text-to-video

curl https://moonnexai.com/v1/videos \
  -H "Authorization: Bearer <MOONNEXAI_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.5-cl",
    "prompt": "清晨海边，镜头平稳向前推进，电影感光线。",
    "duration": 4,
    "aspect_ratio": "16:9",
    "size": "480p",
    "generate_audio": true
  }'

Picture reference

references[] is recommended for new access. Pictures can be a stable public network URL that can be retrieved without going in or Asset:// media assets quotes that are ready to be completed by pressing seedance-2.5-cl:

{
  "model": "seedance-2.5-cl",
  "prompt": "图中女孩对着镜头说“茄子”，镜头缓慢环绕。",
  "references": [
    {
      "media_type": "image",
      "role": "reference_image",
      "url": "Asset://asset_xxx",
      "alias": "人物"
    }
  ],
  "duration": 4,
  "aspect_ratio": "9:16",
  "size": "480p",
  "generate_audio": true
}

Official compatibility parameters

seedance-2.5-cl accepts Volcengine-compatible content[], ratio, resolution, duration and generate_audio on POST /v1/videos. Use POST /api/v3/contents/generations/tasks to retain the compatible path. Both endpoints require a MoonNexAI API key and model: seedance-2.5-cl.

Compatibility describes the request format. It does not establish support for every advanced parameter or arbitrary value. Follow this page's duration, resolution, media-count and role limits.

MoonNexAI Common	Official Compatibility
prompt	type=text entries in content[]
references[]	image_url, video_url, audio_url entries in content[]
aspect_ratio	ratio
size	resolution
duration, generate_audio	field is the same; visible generate_audio=false can close track

The new access is still recommended references[]. The same request selects an media assets expression, not to submit content[] and references[] at the same time, and not to repeat prompt and content[].text.

The following is the official request for authenticating the picture, and the media assets in the example must have been prepared by pressing seedance-2.5-cl:

curl https://moonnexai.com/v1/videos \
  -H "Authorization: Bearer <MOONNEXAI_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.5-cl",
    "content": [
      { "type": "text", "text": "图中女孩对着镜头说“茄子”，360度环绕运镜" },
      {
        "type": "image_url",
        "role": "reference_image",
        "alias": "Image1",
        "image_url": { "url": "Asset://asset_xxx" }
      }
    ],
    "duration": 4,
    "ratio": "9:16",
    "resolution": "480p",
    "generate_audio": true,
    "output_format": "mp4"
  }'

When using the official compatible access, the request path in the previous example is replaced by /api/v3/contents/generations/tasks,request body. Asset:// in content[].image_url.url has the same authentication media assets syntax as references[].url; the error is reported directly by the error resolution failure and is not converted to a public network URL. media assets certification is still subject to content security clearance at the time of generation.

Authenticate media assets

To reuse authenticated assets, call POST /v1/assets/uploads with mode=asset and model=seedance-2.5-cl, then wait for asset details to show generation_ready=true and preparation_status=ready. Explicitly pass the returned Asset://asset_xxx when creating the video.

Authenticate media assets references and common public network URLs are two different input syntax:

• Asset://asset_xxx indicates the use of media assets, which has been prepared and certified according to the current model.
• Video request returns the error directly if the media assets quotes the available binding, state is missing or the reference resolution failed.
• The authentication of media assets does not downgrade to a public network URL; if the operation requires the use of a public network URL, it should be explicitly uploaded into the public network URL in request.
• Asset:// media assets of the different Seedance model line is not directly mixed and must be re-prepared or confirmed with the actual generation model.

Create an authenticated asset

curl https://moonnexai.com/v1/assets/uploads \
  -H "Authorization: Bearer <MOONNEXAI_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.5-cl",
    "mode": "asset",
    "type": "image",
    "url": "https://example.com/person.png"
  }'

media assets upload and authentication itself does not charge video generation costs; a follow-up video task uses model billing. media assets is not automatically attached to new task, and request is visible to media assets references at each time.

Mixed media references

{
  "model": "seedance-2.5-cl",
  "prompt": "保持 @人物 的外观，参考 @动作 的镜头节奏，并融合 @音乐。",
  "references": [
    { "media_type": "image", "role": "reference_image", "url": "Asset://asset_image_xxx", "alias": "人物" },
    { "media_type": "video", "role": "reference_video", "url": "Asset://asset_video_xxx", "alias": "动作" },
    { "media_type": "audio", "role": "reference_audio", "url": "Asset://asset_audio_xxx", "alias": "音乐" }
  ],
  "duration": 8,
  "aspect_ratio": "16:9",
  "size": "720p",
  "generate_audio": true
}

Audio references must be submitted with a picture or video reference, and cannot be submitted separately. Each media assets role should be clearly filled in, and alias should be unique in the same request.

Parameters

field	Type	Annotations
model	string	It's gonna be filled, seedance-2.5-cl.
prompt	string	Video description, with @alias for reference media assets.
references[]	array	Picture, video or audio references; entries are media_type, role, url and optional alias.
content[]	array	Officially compatible text and media arrays; with references[] and two, for full examples see above.
duration	integer	4 - 30 sec.
aspect_ratio / ratio	string	Image aspect ratio; specific usable values are based on modelling capabilities.
size / resolution	string	480p, 720p or 1080p.
generate_audio	boolean	Whether or not to generate a track, default true when omitted.
watermark	boolean	Add watermarks.
callback_url	string	Optional callback address.

Query Results

Create response has been submitted for task. Save id or task_id and continue searching for task; only when status=succeeded is available will the final video address be read. Read error if it fails, do not treat URLs from failed response.

curl https://moonnexai.com/v1/videos/{task_id} \
  -H "Authorization: Bearer <MOONNEXAI_API_KEY>"

Relevant Pages

• Seedance-2 Video Overview
• Video Model Matrix
• Create Video Job
• Uploading and Media Asset Overview