Seedance 2.5 Flash API for 480p and 720p Video

Overview

Expected API availability: November 15, 2026 at 06:00 UTC (14:00 UTC+8). Before launch, generation requests return HTTP 404 with Model unavailable.

Seedance 2.5 Flash generates 480p and 720p videos from text, frame images, or image, video, and audio references. Use model ID seedance-2-5-flash.

Endpoints and authentication

Use the base URL https://api.apixo.ai/api/v1 and send Authorization: Bearer YOUR_API_KEY on each request.

Method	Endpoint	Purpose
POST	`/generateTask/seedance-2-5-flash`	Submit a generation task
GET	`/statusTask/seedance-2-5-flash?taskId={taskId}`	Query status and retrieve the result

Create a video

curl -X POST "https://api.apixo.ai/api/v1/generateTask/seedance-2-5-flash" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "mode": "text-to-video",
      "prompt": "A red paper boat floats on calm blue water. Static camera.",
      "resolution": "480p",
      "duration": 5,
      "aspect_ratio": "16:9"
    }
  }'

Parameters

Field	Required	Default	Description
`request_type`	No	`async`	Use `async` for polling or `callback` for webhook delivery.
`callback_url`	For callbacks		Public HTTPS URL that receives the final result.
`input`	Yes		Object containing the generation parameters below.
`input.mode`	Yes		`text-to-video`, `first_and_last_frames`, or `omni_reference`.
`input.prompt`	Yes		Non-empty text, up to 10,000 characters.
`input.resolution`	No	`720p`	`480p` or `720p`.
`input.duration`	No	`5`	An integer from 4 to 30 seconds, or `-1` for automatic duration. Numeric strings such as `"5"` are accepted.
`input.aspect_ratio`	No	`auto`	`auto`, `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, or `21:9`.
`input.sound`	No	`true`	Generate synchronized audio.
`input.watermark`	No	`false`	Add a visible watermark.
`input.image_urls`	In frame mode		Up to two URLs in frame mode or up to ten in reference mode.
`input.video_urls`	No		Up to five reference video URLs in reference mode.
`input.audio_urls`	No		Up to five reference audio URLs in reference mode.

Mode requirements

Mode	Media requirements
`text-to-video`	A text prompt.
`first_and_last_frames`	One image for the first frame or two images in first-to-last order. Set `aspect_ratio` to `auto` or omit it.
`omni_reference`	At least one image, video, or audio URL. Each reference video or audio clip must be 2–30 seconds. The combined duration is limited to 30 seconds for video references and 30 seconds for audio references.

Retrieve the result

Task creation returns data.taskId. Poll the status endpoint with that ID until data.state is success or failed. For webhook delivery, send request_type: "callback" with callback_url. See Webhooks. On success, parse data.resultJson as JSON and read resultUrls for the video URL. On failure, read data.failCode and data.failMsg.

Billing

Resolution	Without reference video	With reference video
480p	$0.092 per output second	$0.057 per billable second
720p	$0.207 per output second	$0.128 per billable second

Without reference video. Bill the output duration.
With reference video. Bill the output duration plus the greater of the combined reference video duration and two-thirds of the output duration rounded up to a whole second. Only omni_reference requests with video_urls use this rate.
Automatic duration. The initial charge uses 30 output seconds and the applicable reference video charge. After success, any excess is refunded using the actual output duration, capped at 30 billable output seconds.

The request total is rounded up to the nearest $0.001 after applicable discounts.

MCP integration

The model schema declares video decoding in mcp.result and file bindings for image_urls, video_urls, and audio_urls in mcp.attachments. Preserve the order of frame images.