Seedance task failed with a content moderation error: what gets rejected and how to rewrite prompts

When a Seedance task fails with a content moderation error, the request is rejected before a video is produced, and you are not charged for it. According to the official relaydance.com docs, 「Failed or errored requests are never billed」. To recover, poll the task status, read the failed state, then rewrite your prompt to remove the flagged content and resubmit through the OpenAI-compatible endpoint at https://relaydance.com/v1.

What happens when moderation rejects a task

A rejected task ends in a failed status and carries no charge to your account. You submit with POST /v1/video/generations, then poll GET /v1/video/generations/{task_id} until the status is succeeded or failed. On rejection the status returns failed and no video url is produced. Because billing is pay-as-you-go per generated video, a moderation failure costs nothing: as stated in the official relaydance.com docs, 「Failed or errored requests are never billed」. This applies across Seedance 2.0 tiers, including 720p at about $0.190 per second and 1080p at about $0.470 per second (see relaydance.com/models).

How to rewrite a prompt and resubmit

To recover, edit the prompt and reference media, then send a fresh request through the same OpenAI-compatible interface. Follow these steps:

1. Read the failed task response from GET /v1/video/generations/{task_id}.
2. Remove or reword the flagged content in your prompt.
3. Check metadata.content[] reference media and their @image1 to @imageN citations (up to 9 reference images, 3 reference videos and 3 audio tracks per request).
4. Resubmit with POST /v1/video/generations using your model, prompt, and seconds (clips up to 15 seconds).
5. Poll again, or set metadata.callback_url for webhook delivery.

According to the official relaydance.com/docs, 「Set base_url to https://relaydance.com/v1 and keep your OpenAI SDK」, so no client rewrite is needed.

Cost impact of retries across models

Retries after a moderation failure add no cost for the failed attempt, and you pay only for a successful generation. Live per-model rates are published at relaydance.com/models. The table below lists example rates so you can plan a successful resubmission.

For image tasks with gpt-image-2, image output is free and only input is billed, with image-to-image starting from about CNY 0.035, and 4K and 1K output cost the same.

Authentication and payment for resubmitted tasks

Resubmissions use the same authentication and billing setup as the original request. Send Authorization: Bearer YOUR_API_KEY with each call, and create keys at https://relaydance.com/console. Payments are supported through USDT and Stripe card, and billing stays pay-as-you-go per generated video. Because a moderation failure is never billed, iterating on prompts until a task reaches succeeded status only charges you for the clip that completes. Direct OpenAI-compatible access to Seedance requires no BytePlus enterprise account and no KYC, so you can point your existing OpenAI SDK at https://relaydance.com/v1 (see relaydance.com/docs) and retry immediately.

FAQ

Am I charged when a task fails moderation? No. Failed or errored requests are never billed, so a moderation rejection costs nothing.

How do I know a task was rejected? Poll GET /v1/video/generations/{task_id} until the status is succeeded or failed; a rejected task returns failed with no video url.

Do I need to change my code to resubmit? No. Set base_url to https://relaydance.com/v1, keep your OpenAI SDK, edit the prompt, and send POST /v1/video/generations again.