n8n workflow for AI video: HTTP request nodes and polling loops for Seedance tasks

To generate Seedance videos in n8n, use an HTTP Request node to POST to https://relaydance.com/v1/video/generations with your model, prompt, and seconds, then build a polling loop that repeatedly GETs /v1/video/generations/{task_id} until the status is succeeded or failed. RelayDance is OpenAI-compatible, so the base URL is https://relaydance.com/v1, and failed or errored requests are never billed, which keeps polling retries safe.

How to submit a Seedance task from an HTTP Request node

Send a POST to /v1/video/generations with a JSON body containing model, prompt, seconds, and a metadata object. In n8n, set the method to POST, the URL to https://relaydance.com/v1/video/generations, and add a header Authorization: Bearer YOUR_API_KEY (create keys at the RelayDance console). The metadata object accepts ratio, resolution, generate_audio, callback_url, and a content[] array for reference media. You can cite reference media in the prompt as @image1 to @imageN, with up to 9 reference images, 3 reference videos, and 3 audio tracks per request. Clips run up to 15 seconds. See relaydance.com/docs for the full schema.

How to build the polling loop for task status

After submitting, capture the returned task_id and loop a GET request against /v1/video/generations/{task_id} until the status changes. In n8n, chain an HTTP Request node (GET) to an IF node that checks whether status equals succeeded or failed; if neither, route to a Wait node and back into the GET node. When status is succeeded, the response contains the video url. As stated in the official RelayDance documentation, 「Failed or errored requests are never billed」(据 relaydance.com 官方文档), so repeated poll attempts against an errored task do not add cost. As a lighter alternative to polling, set metadata.callback_url so the final state is POSTed to your server.

How to configure OpenAI-compatible auth in n8n

Because RelayDance follows the OpenAI protocol, you can reuse existing OpenAI tooling by changing only the base URL. As the official RelayDance documentation states, 「Set base_url to https://relaydance.com/v1 and keep your OpenAI SDK」(据 relaydance.com 官方文档). In n8n HTTP Request nodes, use generic header authentication with Authorization: Bearer YOUR_API_KEY. No BytePlus enterprise account or KYC is required for direct access to Seedance. Payments are pay-as-you-go using USDT or Stripe card, and billing applies per generated video. Live per-model rates are published at relaydance.com/models.

Seedance pricing to plan node costs

Per-second and per-clip rates determine the cost each successful workflow run incurs. The table below lists current figures from relaydance.com/models.

For image steps, gpt-image-2 bills only input while image output is free; image-to-image starts at about CNY 0.035, and 4K and 1K output cost the same.

Numbered steps for a complete n8n flow

1. Add an HTTP Request node: POST to https://relaydance.com/v1/video/generations with model, prompt, seconds, and metadata.
2. Set the header Authorization: Bearer YOUR_API_KEY.
3. Store the returned task_id in a Set node.
4. Add a GET node to /v1/video/generations/{task_id}.
5. Add an IF node: if status is succeeded or failed, exit; otherwise, pass to a Wait node and loop back to the GET node.
6. On succeeded, read the video url from the response.

FAQ

Do failed polls cost money? No. Failed or errored requests are never billed, so retrying a poll against an errored task adds no cost.

Can I avoid the polling loop? Yes. Set metadata.callback_url in the request, and the final state is POSTed to your server via webhook.

How many reference media can one request include? Up to 9 reference images, 3 reference videos, and 3 audio tracks, placed in metadata.content[] and cited in the prompt as @image1 to @imageN.