Skip to main content
Krea provides API access to ByteDance’s Seedance video generation models. Use the Seedance API on Krea to generate videos from text, animate images, or guide generation with image, video, and audio references. Requests use a Krea API token and the https://api.krea.ai base URL.

​
Available Seedance API models

These models support text-to-video, start and end frames, reference media, and native audio. See the API reference for each model’s current request schema. Check API pricing ↗ before submitting a job; price depends on the model and generation settings.
Krea’s Seedance 2.0 and 2.0 Fast endpoints include optional Krea prompt enhancement and upscaling beyond the base model. Seedance 2.5 includes optional Krea prompt enhancement. For Seedance 2.5 video-editing requests with reference_videos, the output length and aspect ratio come from the source video; duration and aspect_ratio are ignored for those requests.

​
Get a Seedance API key

Create a Krea API token and add API balance using API keys and billing. Store the token in the KREA_API_KEY environment variable on your server. Send it as Authorization: Bearer $KREA_API_KEY; never expose it in browser code. Krea app compute and API balance are separate. Seedance requests through the Krea API use your workspace’s API balance.

​
Generate a video

Submit a prompt to the Seedance 2.5 endpoint for text-to-video generation. These examples request a five-second, 720p landscape video.
Install the JavaScript SDK with npm install @krea-ai/sdk. The REST request returns a job_id, not the finished video. Poll GET /jobs/{job_id} every 2–5 seconds until status is completed, then read the video URL from result.urls[0]. Stop polling on failed or cancelled. The SDK’s jobs.wait method handles polling for you.
cURL
For production integrations, use webhooks to receive completion events. See job lifecycle and errors and debugging for job states and failure handling.

​
Animate an image

For Seedance image-to-video generation, use the same endpoint and add start_image to your request. Pass a publicly reachable image URL, an uploaded asset URL, or a base64 data URI.
Replace start_image with your own image URL. To control the final frame, add end_image. For reference-guided generation, use reference_images, reference_videos, and reference_audios according to the selected model’s API schema.

​
Frequently asked questions

​
Does Krea offer a ByteDance Seedance API?

Yes. Krea offers ByteDance Seedance 2.5, Seedance 2.0, Seedance 2.0 Fast, and Seedance 2.0 Mini through its REST API at https://api.krea.ai. Authenticate with a Krea API token and submit a request to a Seedance endpoint listed above.

​
Can I use Seedance for both text-to-video and image-to-video?

Yes. Send a prompt for text-to-video, or add start_image to animate an image. Both workflows use the same Seedance model endpoint and return an asynchronous job.

​
Do I need a separate ByteDance API key?

No. Calls to Seedance through Krea use your Krea API token and workspace API balance. You do not need to send a separate ByteDance credential with these requests.

​
How much does the Seedance API cost on Krea?

See Krea API pricing ↗ for current Seedance prices and available settings. API usage is billed separately from Krea app compute. Failed and cancelled jobs are not billed.

​
Can I call Seedance from an AI agent?

Yes. Connect an agent to the Krea MCP server to discover available video models and their schemas. You can also call the REST API or JavaScript SDK directly from your application.