Seedance MCP
Generate Seedance actions and video content in the AI Agent through MCP.
ByteDance's strongest video generation model, supporting text and image inputs.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View DocsAccess Seedance 1.x, 2.0, and 2.5 through a unified API. Supports text-to-video, image-to-video, full-modal references, video editing, and extension—up to 4k and 30 seconds, with pay-as-you-go pricing.
Seedance 2.5 extends full-modal video capabilities to up to 30 seconds, allowing combinations of image, video, and audio references, while supporting editing or extending existing videos.
Ace Data Cloud provides a unified Seedance REST API covering multiple model versions, asynchronous task polling, and callback notifications, with no need to switch interfaces for different capabilities.
From text descriptions to high-definition video—Seedance provides comprehensive video generation capabilities
Enter a text description to generate high-quality video. Seedance accurately understands motion semantics, with smooth and natural character movements.
Upload a reference image, and the model generates dynamic video based on its content. Perfectly preserves the characters, scenes, and style in the image.
Supports 480p, 720p, 1080p, and 4k depending on the model. Flexibly choose between fast previews and high-definition final videos.
Enable the generate_audio option, and Seedance will automatically add matching sound effects and background music to generated videos.
Receive a task_id after submitting a video generation request, then retrieve results through polling or callbacks. The polling API is completely free.
Enable return_last_frame to obtain the final frame image of a video, making it convenient for continued video generation or previewing.
import requests
# Step 1: Submit video generation task
response = requests.post(
"https://api.acedata.cloud/seedance/videos",
headers={
"Authorization": "Bearer YOUR_API_KEY",
"Content-Type": "application/json"
},
json={
"model": "doubao-seedance-1-5-pro-251215",
"content": [
{"type": "text", "text": "A cat playing piano in a cozy room"}
],
"generate_audio": True
}
)
task_id = response.json()["id"]
# Step 2: Poll for result (free)
result = requests.get(
f"https://api.acedata.cloud/seedance/tasks/{task_id}",
headers={"Authorization": "Bearer YOUR_API_KEY"}
)
print(result.json()["data"]["video_url"])
Submit a text description and retrieve video results asynchronously—a clean RESTful API design.
POST to /seedance/videos, pass text or image content, and receive a task_id
Query /seedance/tasks with the task_id, then retrieve the video URL once the status is complete
Set callback_url, and a POST notification will be sent to your server when generation is complete
From creative short films to commercial ads—Seedance is being used in these scenarios
Quickly generate creative short-form video content for platforms such as TikTok / Douyin / YouTube Shorts
Generate product advertising videos in bulk, significantly reducing video production costs and improving ad asset output efficiency
Automatically convert text lesson plans into animated videos for a more intuitive learning experience
Game cutscenes and concept video prototypes—quickly validate ideas and reduce pre-production costs
From registration to generating your first video, it takes only a few minutes
Create a free account on Ace Data Cloud and generate your Bearer Token from the console.
POST to /seedance/videos, provide a text description or reference image, and choose a model and resolution.
Retrieve the completed video URL through polling or a callback, then download and use it.
From quick previews to high-quality final videos—multiple models cover different needs
A full-modal video model with up to 30 seconds and 1080p, supporting more reference assets, audio-only references, editing, and extension.
The fastest and lowest-cost Pro model. Ideal for quick previews and batch generation.
A lightweight model for specific scenarios, with two variants: T2V (text-to-video) and I2V (image-to-video).
Billed by video duration (seconds); pricing depends on the model, resolution, and reference materials. USD estimates are calculated based on Seedance's current largest package $13,584 / 142,800 Credits (approximately $0.095126/Credit); final pricing is subject to the real-time pricing page. The polling API is free.
| Model | 480p | 720p | 1080p |
|---|---|---|---|
| Seedance 2.5 | Approx. $0.113/sec | Approx. $0.243/sec | Approx. $0.605/sec |
| Seedance 2.5 + Reference Video | Approx. $0.273/sec | Approx. $0.586/sec | Approx. $1.444/sec |
| Seedance 2.0 Standard | Approx. $0.074/sec | Approx. $0.160/sec | Approx. $0.400/sec |
| Seedance 2.0 Standard + Reference Video | Approx. $0.182/sec | Approx. $0.391/sec | Approx. $0.973/sec |
| Seedance 2.0 Fast | Approx. $0.060/sec | Approx. $0.129/sec | — |
| Seedance 2.0 Fast + Reference Video | Approx. $0.143/sec | Approx. $0.306/sec | — |
| Seedance 2.0 Mini | Approx. $0.037/sec | Approx. $0.081/sec | — |
| Seedance 2.0 Mini + Reference Video | Approx. $0.090/sec | Approx. $0.195/sec | — |
| Seedance 1.5 Pro | Approx. $0.013/sec | Approx. $0.028/sec | Approx. $0.063/sec |
| Seedance 1.5 Pro + Audio | Approx. $0.026/sec | Approx. $0.056/sec | Approx. $0.126/sec |
| Seedance 1.0 Pro | Approx. $0.024/sec | Approx. $0.054/sec | Approx. $0.118/sec |
| Seedance 1.0 Pro Fast | Approx. $0.007/second | Approx. $0.015/second | Approx. $0.033/second |
| Seedance 1.0 Lite | Approx. $0.016/second | Approx. $0.036/second | Approx. $0.079/second |
The task status polling endpoint /seedance/tasks is completely free
Everything you need to know about using the Seedance API
Seedance supports generation from text, images, video, and audio references. 2.0 Standard supports up to 4k, while 2.5 supports up to 30 seconds and can edit or extend existing videos.
Yes! Include content of the image_url type in the content array to enable image-to-video (I2V). The model preserves character appearance, scene layout, and artistic style from the image while adding natural motion effects.
Generation time depends on the model and resolution. Pro Fast models typically finish within 1-2 minutes, while standard Pro models and high-resolution videos may take 3-5 minutes. We recommend using callback notifications (callback_url) to avoid frequent polling.
No. The /seedance/tasks endpoint is completely free, and you can poll task status as often as you like until video generation is complete. Charges only apply when submitting a /seedance/videos generation request.
Choose Seedance 2.5 when you need up to 30 seconds, more reference materials, or editing/extending; choose 2.0 Standard when you need up to 4k; choose 2.0 Fast or Mini for quick previews and batch generation.
Set generate_audio: true in supported 1.5 Pro and 2.x models to include matching sound in the video. Specific charges are subject to the rules displayed on the pricing page by model and resolution.
We offer a variety of AI creative tools—from video to images to music
Generate high-quality AI videos with simple API calls. Pay as you go—no subscription fees, no commitments.
ByteDance Seedance video generation API allows you to input images and text to generate videos.
ByteDance Seedance Task Query API
We offer various credit packages. The more you purchase, the higher the discount you receive.
up to46% OFFGenerate Seedance actions and video content in the AI Agent through MCP.
Configuration Method
Connect to the current service in Claude Code:
/mcp to check the connection and tool list.export ACEDATACLOUD_API_TOKEN='YOUR_API_TOKEN'
claude mcp add --scope user --transport http 'seedance-mcp' 'https://seedance.mcp.acedata.cloud/mcp' --header 'Authorization: Bearer ${ACEDATACLOUD_API_TOKEN}'
Provide prompt words, material input, and task processing guidelines for AI Agent regarding ByteDance Seedance video and action generation.
Install the current Skill to Claude Code. Project-level configurations are shared across the repository, while user-level configurations apply to all projects on this machine.
npx skills add acedatacloud/skills --skill 'seedance-video' -a 'claude-code' --yes