Text to video API quickstart with an OpenAI compatible endpoint

RelayDance provides an OpenAI-compatible text-to-video API at https://relaydance.com/v1 supporting ByteDance Seedance 2.0 and 1.5 Pro, HappyHorse video models, and Grok Imagine video and image models. Submit POST requests with your prompt, duration, and model selection, then poll the task status until generation completes. Authentication uses bearer tokens created in your console, and pricing is 30 percent or more below official rates with no charges for failed requests.

Authentication and setup

Access RelayDance's API by creating an API key at https://relaydance.com/console. Include your key in all requests using the Authorization header format: Authorization: Bearer YOUR_API_KEY. The base URL for all endpoints is https://relaydance.com/v1. You can pay via USDT or Stripe card. Check live per-model pricing at https://relaydance.com/models to see current rates, which are maintained 30 percent or more below official vendor prices. Failed or errored requests incur no charges.

Video generation request and polling

To generate a video, POST to /v1/video/generations with your model name, text prompt, duration in seconds, and optional metadata for aspect ratio, resolution, audio generation, and callback URL. The API returns a task ID immediately. Poll GET /v1/video/generations/{task_id} until the status field shows succeeded or failed. The completed response includes the video URL ready for download. Alternatively, set metadata.callback_url to enable webhook mode: the final state will be POSTed to your server when ready, eliminating the need for polling.

Reference media and multimodal prompts

RelayDance supports reference media to guide video generation: up to 9 images, 3 video clips (max 15 seconds each), and 3 audio tracks per request. Include references in the metadata.content[] array and cite them in your prompt using @image1, @image2, @imageN syntax for images, with analogous references for videos and audio. This allows you to blend your text description with visual and audio context for more precise output control and creative direction in the generated video.

Image generation

Beyond video, RelayDance also provides image generation via POST /v1/images/generations. Grok Imagine and other supported models generate images from text prompts using the same authentication method and OpenAI-compatible request format. Image results are returned in the response and can serve as reference media for subsequent video generation tasks, enabling creative workflows that combine image and video synthesis within a single API platform.

Quick reference: video generation workflow

1. Create an API key at https://relaydance.com/console and note your Bearer token
2. POST to /v1/video/generations with model, prompt, seconds, and metadata fields
3. Capture the task_id from the response
4. Poll GET /v1/video/generations/{task_id} every 5-10 seconds, or configure callback_url for webhook notifications
5. When status equals succeeded, retrieve your video URL and download the file
6. Verify charges only applied to successful generations via your console billing

FAQ

• Which video models does RelayDance support? RelayDance offers ByteDance Seedance 2.0 and Seedance 1.5 Pro, HappyHorse video models, and Grok Imagine video models. See https://relaydance.com/models for current availability and pricing.
• How do I cite reference images in my prompt? Include reference media in metadata.content[] and reference each item by position in your prompt text: @image1, @image2, @video1, @audio1, etc., up to nine images, three videos, and three audio tracks per request.
• Am I charged for failed video generations? No. RelayDance only bills successful requests. Failed or errored tasks incur no cost, allowing safe experimentation with different prompts and parameters.