> ## Documentation Index > Fetch the complete documentation index at: https://docs.haitoken.ai/llms.txt > Use this file to discover all available pages before exploring further. # Seedance Video Generation > Video generation API compatible with the official Seedance protocol, supporting text-to-video, image-to-video, reference media, and synchronized audio This endpoint is fully compatible with the official Seedance protocol, so you can reuse the parameter structure from the official documentation. `model` must be a platform model identifier, which the gateway rewrites to the channel model name after routing. `callback_url` is overridden and injected by the gateway with a signed default callback address. ## Features * The `content` element list supports five types: text, image, video, audio, and draft task * Supports first frame / last frame / reference images (1-9 for the 2.0 series) / reference video / reference audio * Supports synchronized audio via `generate_audio` (Seedance 2.0 series and 1.5 Pro only) * Supports online/offline inference via `service_tier` and task timeout via `execution_expires_after` * Supports draft mode `draft` (1.5 Pro only), last frame return `return_last_frame`, and task priority `priority` (2.0 only) * Supports video editing (2.5 only), edit the visuals or audio of the original video, adding, deleting, or modifying objects in the video, or redrawing or repairing part of the frame. * Supports Video extension (2.5 only), Extend the original video forward or backward. ## Authentication Include the `Authorization` header in the format `Bearer YOUR_API_KEY`. ## Supported Video Models | Model Identifier | Model Type | Model Description | | ----------------- | --------------- | --------------------------------------------------------------------------------------------------------------------------------------------- | | seedance-2.5 | Reference Video | seedance 2.5 is a next-generation video creation model with major improvements in long-form storytelling, multimodal referencing, and editing | | seedance-2.0 | Reference Video | seedance 2.0 core video generation model | | seedance-2.0-mini | Reference Video | new generation high-cost-benefit video generation model | | seedance-2.0-fast | Reference Video | inherited the core functions and advantages of seedance-2.0, faster speed | ## Quick example ```python Python theme={null} import requests url = "https://api.haitoken.ai/v1/seedance/video/generations" headers = { "Authorization": "Bearer YOUR_API_KEY", "Content-Type": "application/json" } data = { "model": "seedance-2.0", "content": [ {"type": "text", "text": "A golden retriever runs along the beach at sunset, camera slowly tracking"}, { "type": "image_url", "image_url": {"url": "https://example.com/images/dog.png"}, "role": "first_frame" } ], "duration": 5, "resolution": "720p", "ratio": "16:9", "generate_audio": True, "watermark": False } response = requests.post(url, json=data, headers=headers) print(response.json()) ``` ```javascript Node.js theme={null} const response = await fetch('https://api.haitoken.ai/v1/seedance/video/generations', { method: 'POST', headers: { 'Authorization': 'Bearer YOUR_API_KEY', 'Content-Type': 'application/json' }, body: JSON.stringify({ model: 'seedance-2.0', content: [ { type: 'text', text: 'A golden retriever runs along the beach at sunset, camera slowly tracking' }, { type: 'image_url', image_url: { url: 'https://example.com/images/dog.png' }, role: 'first_frame' } ], duration: 5, resolution: '720p', ratio: '16:9', generate_audio: true, watermark: false }) }); const data = await response.json(); console.log(data); ``` ```curl cURL theme={null} curl -X POST 'https://api.haitoken.ai/v1/seedance/video/generations' \ -H 'Authorization: Bearer YOUR_API_KEY' \ -H 'Content-Type: application/json' \ -d '{ "model": "seedance-2.0", "content": [ {"type": "text", "text": "A golden retriever runs along the beach at sunset, camera slowly tracking"}, { "type": "image_url", "image_url": {"url": "https://example.com/images/dog.png"}, "role": "first_frame" } ], "duration": 5, "resolution": "720p", "ratio": "16:9", "generate_audio": true, "watermark": false }' ``` A successful creation returns the task ID: ```json theme={null} { "id": "cgt-20260730120000-a1b2c3" } ``` ## Next steps * See [Query Seedance video task](/docs/en/api-reference/video/seedance-status) to retrieve generation results * See [Video generation](/docs/en/api-reference/video/generation) for the unified protocol * See the [Wan protocol](/docs/en/api-reference/video/wan) for native Alibaba Wan parameters * See the [model list](/docs/en/api-reference/models/list-models) for available video models ## OpenAPI ````yaml en/api-reference/video/seedance/openapi.json POST /v1/seedance/video/generations openapi: 3.0.1 info: title: Default Module description: '' version: 1.0.0 servers: [] security: [] tags: [] paths: /v1/seedance/video/generations: post: tags: [] summary: Create video generation task (seedance compatible protocol) description: >-

Body is identical to the official seedance API; model must be a platform model identifier (rewritten to the channel model name after routing), callback_url is overridden and injected by the gateway (with signature). parameters: - name: Authorization in: header description: API Key token (Bearer sk-xxx) required: true example: 'Bearer ' schema: type: string - name: Content-Type in: header description: '' example: application/json schema: type: string requestBody: content: application/json: schema: $ref: '#/components/schemas/SeedanceCreateTaskRequest' description: '' examples: {} responses: '200': description: '' content: application/json: schema: $ref: '#/components/schemas/SeedanceCreateTaskResponse' headers: {} deprecated: false security: [] components: schemas: SeedanceCreateTaskRequest: type: object properties: callback_url: type: string description: >- Result callback URL (the upstream POSTs notifications on task status changes; when the gateway has a callback configured, it overrides and injects a signed default address) return_last_frame: type: boolean description: >- Whether to return the last frame image of the generated video (PNG, same resolution as the video, no watermark), default false; can be used as the first frame of the next task to quickly generate multiple consecutive video segments service_tier: type: string description: >- Service tier: default online inference (default) / flex offline inference (50% of the online price); the Seedance 2.0 series only supports online inference and does not support this parameter execution_expires_after: type: integer description: >- Task timeout threshold (seconds), counted from created_at, range [3600, 259200], default 172800 (48 hours); the task is automatically terminated and marked as expired after timeout camera_fixed: type: boolean description: >- Whether to fix the camera, default false; not supported in reference image scenarios and the Seedance 2.0 series generate_audio: type: boolean description: >- Whether to generate audio synchronized with the visuals (voice/sound effects/background music), default true; only supported by the Seedance 2.0 series and 1.5 Pro safety_identifier: type: string description: >- Unique end-user identifier (for platform security audit, English only, max 64 characters; it is recommended to pass a hash of the username/user ID/email to avoid leaking privacy) model: type: string description: >- Platform model identifier (e.g. seedance-2.0), rewritten to the channel model name by the gateway after routing content: type: array items: $ref: '#/components/schemas/ContentItem' description: Content element description: >- Content element list (text prompt / image / video / audio / draft task ID) resolution: type: string description: >- Generated video resolution: 480p / 720p / 1080p / 4k; the Seedance 2.0 series and 1.5 Pro default to 720p, 1.0 Pro / Pro Fast default to 1080p; 1080p is not supported by 2.0 Fast / 2.0 Mini, 4k is only supported by Seedance 2.0 ratio: type: string description: >- Generated video aspect ratio: 16:9 / 4:3 / 1:1 / 3:4 / 9:16 / 21:9 / adaptive (intelligently selected based on the input) duration: type: integer description: >- Generated video duration (integer seconds); choose either duration or frames, frames takes precedence, default 5; Seedance 1.0 Pro / Pro Fast: [2, 12]; 1.5 Pro: [4, 12] or -1 (model intelligently selects the duration); Seedance 2.0 series: [4, 15] or -1 frames: type: integer description: >- Generated video frame count (choose either duration or frames, frames takes precedence), used to generate fractional-second durations; only supported by Seedance 1.0 Pro / Pro Fast (not supported by the 2.0 series and 1.5 Pro), range [29, 289] and must satisfy 25 + 4n (frame rate 24, frames = seconds x 24) seed: type: integer description: >- Random seed, range [-1, 2^32-1], default -1 (random); not supported by the Seedance 2.0 series format: int64 watermark: type: boolean description: >- Whether the generated video carries an "AI Generated" watermark, default false draft: type: boolean description: >- Whether to enable draft mode (480p preview video at a lower cost for quick validation), default false; only supported by Seedance 1.5 Pro; draft mode does not support last frame return or offline inference priority: type: integer description: >- Task execution priority 0-9, higher values queue earlier, default 0; only supported by Seedance 2.0, only affects the queue order of the same Endpoint, not supported for offline inference (flex) required: - model SeedanceCreateTaskResponse: type: object properties: id: type: string description: Task ID ContentItem: type: object properties: image_url: $ref: '#/components/schemas/MediaRef' description: >- Image input (type=image_url, url supports public URL / base64 data URI / asset://) video_url: $ref: '#/components/schemas/MediaRef' description: >- Video input (type=video_url, only supported by the Seedance 2.0 series; url supports public URL / asset://) audio_url: $ref: '#/components/schemas/MediaRef' description: >- Audio input (type=audio_url, only supported by the Seedance 2.0 series; audio cannot be provided alone, at least 1 reference video or image is required) draft_task: $ref: '#/components/schemas/DraftTaskRef' description: >- Draft task input (type=draft_task, only supported by Seedance 1.5 Pro; reuses the draft task's inputs to generate the final video) type: type: string description: 'Element type: text / image_url / video_url / audio_url / draft_task' text: type: string description: Text prompt (type=text, recommended within 1000 words) role: type: string description: >- Position/purpose of the media element (conditionally required): first_frame / last_frame / reference_image (1-9 images for the 2.0 series) / reference_video / reference_audio required: - type MediaRef: type: object properties: url: type: string description: >- Public media URL / base64 data URI (base64 is stored to OBS by the gateway and replaced) / asset:// required: - url DraftTaskRef: type: object properties: id: type: string description: >- Draft task ID (ModelArk automatically reuses the draft task's model / text / image_url / generate_audio / seed / ratio / duration / camera_fixed inputs to generate the final video) required: - id ````