Skip to main content
OrcaRouter speaks the Seedance video models on the same submit-and-poll endpoint as Kling. You send model: byteplus/dreamina-seedance-2-0-260128, OrcaRouter routes the request to the upstream Ark /contents/generations/tasks API, and you poll the same task ID back through OrcaRouter once it’s done (typically 30 seconds to 4 minutes depending on duration / resolution / generate_audio).
Currently available model. Only Seedance 2.0 is provisioned right now, under the backend name byteplus/dreamina-seedance-2-0-260128. The capability table below lists the rest of the Seedance family for reference, but they are not yet selectable in the playground or routable through OrcaRouter — use byteplus/dreamina-seedance-2-0-260128 for every request for now.
The submit endpoint POST /v1/video/generations and the fetch endpoint GET /v1/video/generations/{task_id} are shared with Kling. What changes is the request body: Kling uses prompt + image + metadata.{mode, aspect_ratio, image_list, ...}, Seedance uses prompt + metadata.{content[], ratio, duration, generate_audio, watermark, ...}. The prefix on model selects which schema is honored.

​
Models

ModelT2VI2V (first)I2V (first+last)Multimodal ref¹Video edit²Generate audio³DurationAvailable
byteplus/dreamina-seedance-2-0-260128 (2.0)✓✓✓✓ full✓✓4 – 15 s✓
byteplus/seedance-2.0-fast✓✓✓✓ full✓✓4 – 15 splanned
byteplus/seedance-1-5-pro✓✓✓image only✓4 – 12 splanned
byteplus/seedance-1-0-pro✓✓✓image only2 – 12 splanned
byteplus/seedance-1-0-pro-fast✓✓image only2 – 12 splanned
byteplus/seedance-1-0-lite-i2v✓✓image only2 – 12 splanned
byteplus/seedance-1-0-lite-t2v✓image only2 – 12 splanned
¹ Multimodal reference = the metadata.content[] array can carry image_url / video_url / audio_url items with role markers (reference_image / reference_video / reference_audio). “Full” means combinations of image + video + audio are accepted. ² Video editing = pass a video_url content item to apply prompt-driven edits to the source video (subject swap, region inpainting, etc.). ³ Native audio = upstream auto-generates a soundtrack matching the video. Toggle via metadata.generate_audio: true. The submit endpoint is the same for every model — POST /v1/video/generations. What changes is which metadata fields the upstream honors per the table above. See the upstream Seedance capability matrix for the authoritative per-model feature list.

​
Submit a task

Send a POST to /v1/video/generations with model, prompt, and any upstream-specific parameters under metadata: 
curl https://api.orcarouter.ai/v1/video/generations \
  -H "Authorization: Bearer sk-orca-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "byteplus/dreamina-seedance-2-0-260128",
    "prompt": "A girl holding a fox, the girl opens her eyes, looks gently at the camera, the fox hugs affectionately, the camera slowly pulls out, the girl'\''s hair is blown by the wind",
    "metadata": {
      "content": [
        { "type": "image_url", "image_url": { "url": "https://example.com/foxgirl.png" } }
      ],
      "ratio": "16:9",
      "duration": 5,
      "generate_audio": true,
      "watermark": false
    }
  }'
Response carries the task ID (same envelope as Kling — OrcaRouter normalizes across providers): 
{
  "id": "task_9q9oz6tjtgABYWC1QIqoz3sscgVz7ycw",
  "task_id": "task_9q9oz6tjtgABYWC1QIqoz3sscgVz7ycw",
  "object": "video",
  "model": "byteplus/dreamina-seedance-2-0-260128",
  "status": "queued",
  "progress": 0,
  "created_at": 1777975188
}
OrcaRouter wraps your prompt as the text item inside Seedance’s content[] array automatically — you don’t need to pass a {type: "text"} item yourself. Any text item you supply in metadata.content[] is replaced by your top-level prompt. Other content items (image_url, video_url, audio_url) pass through unchanged.

​
Body fields

These fields go inside metadata. Arrange them per the variant tables below.
FieldTypeNotes
contentarrayMultimodal reference items. Each item: {type, image_url? | video_url? | audio_url?, role?}. Skip if pure text-to-video.
ratiostringAspect ratio. 16:9 / 9:16 / 1:1 / 4:3 / 3:4 / 21:9 / adaptive. adaptive infers from input.
durationintegerSeconds. Allowed range depends on model — see the table above.
resolutionstring480p / 720p / 1080p. Default 720p. 1080p only on seedance-2.0 / seedance-2.0-fast / seedance-1-5-pro / seedance-1-0-pro / seedance-1-0-pro-fast.
generate_audiobooleanAuto-generate a synced soundtrack. Default false. Only on seedance-2.0 / 2.0-fast / 1-5-pro.
watermarkbooleanImprint the upstream watermark. Default upstream-defined.
seedintegerRandom seed for reproducibility.
service_tierstringdefault (online) or flex (offline / lower priority, higher quota). Defaults to default.
return_last_framebooleanReturn the final frame as an image alongside the MP4. Default false.
callback_urlstringWebhook URL — receives status changes instead of (or alongside) polling.

​
content[] item shape

Each item in metadata.content is one of four shapes: 
{ "type": "image_url", "image_url": { "url": "https://..." }, "role": "first_frame" }
{ "type": "video_url", "video_url": { "url": "https://..." }, "role": "reference_video" }
{ "type": "audio_url", "audio_url": { "url": "https://..." }, "role": "reference_audio" }
{ "type": "text",      "text": "..." }    // automatically replaced by top-level prompt
role values:
rolePurpose
first_frameAnchor this image as the first frame of the generated video.
end_frameAnchor this image as the last frame (use with first_frame for first+last frame i2v).
reference_imageStyle / subject reference (Multimodal reference variant; can pass multiple).
reference_videoStyle / motion reference, or the source video for editing / extension.
reference_audioBackground music or voice reference (audio-video generation).
Reference items inside the prompt with [Image 1], [Video 1], [Audio 1] syntax. The index matches the array order (1-based, scoped per type).

​
Poll for results

Use the task ID returned at submit time: 
curl https://api.orcarouter.ai/v1/video/generations/task_9q9oz6tjtgABYWC1QIqoz3sscgVz7ycw \
  -H "Authorization: Bearer sk-orca-..."
Response shape is wrapped (identical to Kling): 
{
  "code": "success",
  "message": "",
  "data": {
    "task_id": "task_9q9oz6tjtgABYWC1QIqoz3sscgVz7ycw",
    "status": "SUCCESS",
    "progress": "100%",
    "result_url": "https://ark-content-generation-ap-southeast-1.tos-ap-southeast-1.volces.com/.../video.mp4",
    "submit_time": 1777975188,
    "start_time": 1777975241,
    "finish_time": 1777975277,
    "fail_reason": ""
  }
}
Status values are normalized to uppercase across providers:
StatusUpstream Seedance statusMeaning
NOT_START(transient)Task row created, not yet dispatched
SUBMITTEDqueuedSent to upstream, waiting in the queue
IN_PROGRESSrunningUpstream is rendering
SUCCESSsucceededDone. data.result_url carries the MP4
FAILUREfailedFailed. data.fail_reason has the reason
Progress is a percent string ("50%", "100%"), not an integer. Poll every 5 - 10 seconds. A 5-second 720p clip typically completes in 30 - 60 seconds; 1080p with audio or 15-second / multimodal-reference clips can take 3 - 5 minutes. The result_url is an upstream-signed TOS URL with a short TTL — download or rehost promptly if you need long retention.

​
Endpoint variants

All variants share POST /v1/video/generations. Which Seedance feature path the upstream serves is determined by the metadata.content[] items and role markers — not by URL.

​
Text-to-video

Just model + prompt + optional metadata. No content items means pure text-to-video: 
curl https://api.orcarouter.ai/v1/video/generations \
  -H "Authorization: Bearer sk-orca-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "byteplus/dreamina-seedance-2-0-260128",
    "prompt": "Photorealistic style: under a clear blue sky, a vast expanse of white daisy fields stretches out. The camera gradually zooms in on a single daisy with glistening dewdrops on its petals.",
    "metadata": {
      "ratio": "16:9",
      "duration": 5,
      "watermark": true
    }
  }'

​
Image-to-video — first frame

Pass one image item with role: "first_frame": 
curl https://api.orcarouter.ai/v1/video/generations \
  -H "Authorization: Bearer sk-orca-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "byteplus/dreamina-seedance-2-0-260128",
    "prompt": "the cat starts dancing energetically",
    "metadata": {
      "content": [
        { "type": "image_url", "image_url": { "url": "https://example.com/cat.png" }, "role": "first_frame" }
      ],
      "ratio": "adaptive",
      "duration": 5,
      "generate_audio": true
    }
  }'

​
Image-to-video — first and last frame

Two image items, one each for first_frame and end_frame: 
curl https://api.orcarouter.ai/v1/video/generations \
  -H "Authorization: Bearer sk-orca-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "byteplus/dreamina-seedance-2-0-260128",
    "prompt": "Create a 360-degree orbiting camera shot from start to end frame.",
    "metadata": {
      "content": [
        { "type": "image_url", "image_url": { "url": "https://example.com/start.jpg" }, "role": "first_frame" },
        { "type": "image_url", "image_url": { "url": "https://example.com/end.jpg" },   "role": "end_frame"   }
      ],
      "ratio": "16:9",
      "duration": 6
    }
  }'

​
Multimodal reference — image + video + audio

Combine reference_image / reference_video / reference_audio items. Reference them in the prompt with [Image N] / [Video N] / [Audio N] indices (1-based, per type): 
curl https://api.orcarouter.ai/v1/video/generations \
  -H "Authorization: Bearer sk-orca-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "byteplus/dreamina-seedance-2-0-260128",
    "prompt": "Use the first-person POV framing from [Video 1] throughout, and use [Audio 1] as the background music. First-person POV fruit tea promotional ad: [Image 1] hands pick a dew-covered apple; [Image 2] holds the finished drink up to the camera.",
    "metadata": {
      "content": [
        { "type": "image_url", "image_url": { "url": "https://example.com/tea_pic1.jpg" }, "role": "reference_image" },
        { "type": "image_url", "image_url": { "url": "https://example.com/tea_pic2.jpg" }, "role": "reference_image" },
        { "type": "video_url", "video_url": { "url": "https://example.com/tea_video1.mp4" }, "role": "reference_video" },
        { "type": "audio_url", "audio_url": { "url": "https://example.com/tea_audio1.mp3" }, "role": "reference_audio" }
      ],
      "ratio": "16:9",
      "duration": 11,
      "generate_audio": true,
      "watermark": false
    }
  }'
Available on seedance-2.0 and seedance-2.0-fast (full image + video + audio combinations); seedance-1-5-pro and seedance-1-0-* accept only reference_image items.

​
Video editing / extension

Pass {type: "video_url", role: "reference_video"} and ask the prompt to modify or extend it: 
curl https://api.orcarouter.ai/v1/video/generations \
  -H "Authorization: Bearer sk-orca-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "byteplus/dreamina-seedance-2-0-260128",
    "prompt": "Change all the fruits in [Video 1] into fresh fruits.",
    "metadata": {
      "content": [
        { "type": "video_url", "video_url": { "url": "https://example.com/source.mp4" }, "role": "reference_video" }
      ],
      "ratio": "adaptive",
      "duration": 6
    }
  }'
Available on seedance-2.0 and seedance-2.0-fast only.

​
Webhooks

Pass metadata.callback_url: "https://your.domain/webhook" to receive a POST when the task transitions to SUCCESS or FAILURE. The payload mirrors the polling response. If you set both polling and a callback, you’ll get both — they’re independent.

​
Billing

OrcaRouter passes through upstream’s per-task token charge with no markup. Final cost matches ByteDance Ark’s published rate card (the upstream completion_tokens / total_tokens from the task result are converted to quota at the model’s per-token rate set in your Channel Margin config). A small pre-consume hold is reserved at submit time; the difference settles on success. See Operations / Billing & Usage.

​
See also