跳转到主要内容
POST
异步生成 Seedance 2.5 视频。使用统一入口 POST /v1/videos,直接传计费键 sd2.5-480p / sd2.5-720p / sd2.5-1080p。成功返回 task_id,用 查询视频任务 轮询,完成后从 result.video_url 取片。

模型

计费方式为 per_second(按秒)。费用 = 该分辨率单价 × 请求时长。实际单价以 模型目录 与响应中的 cost 为准。

模式判定

本接口不支持首尾帧字段。

参考素材限制

参考素材须为公网可访问的 URL。

请求参数

示例

文生视频

参考图与参考视频

响应

创建成功后返回统一任务结构(含 task_id、status、cost 等)。data.model 为计费键(如 sd2.5-720p)。任务完成后通过查询接口读取 result.video_url。

授权

Authorization
string
header
必填

控制台创建的 API Key,格式:Bearer sk_xxx

请求体

application/json
model
string
必填

omni-10s, veo3.1-fast, veo3.1-lite, veo3.1-lite-hd, grok-imagine-video-1.5 (and fast/lite), minimax-h3, minimax-h3-max, kling-4.0 / kling-4.0-720p / kling-4.0-1080p, sd2.5-480p / sd2.5-720p / sd2.5-1080p, seedance2.0-md, sd2-480p-ac, sd2-720p-ac, sd2-720p-dl, sd2-720p-db, sd2.0-mini-720p, seedance2.5-30s, sd2.5-720p-ac, sd2.5-720p-933-ac, or wan3-480p-20s-ac

示例:

"omni-10s"

prompt
string
必填
duration

omni-10s: fixed 10; seedance2.0-md: 5/10/15; sd2-*-ac: 4–15; sd2-720p-dl: fixed 15; sd2-720p-db: 5/10/15; sd2.0-mini-720p: 4–15; sd2.5-480p / sd2.5-720p / sd2.5-1080p: 5–30; seedance2.5-30s: fixed 30; sd2.5-720p-ac and sd2.5-720p-933-ac: 4–30; wan3-480p-20s-ac: 5–20

seconds

Alias of duration

aspect_ratio
string
ratio
string

Alias of aspect_ratio

orientation
string

portrait/landscape may imply aspect ratio

size
string
resolution
string

omni-10s is fixed at 720p

first_image
string<uri>

First-frame URL for omni-10s frame mode

last_image
string<uri>

Last-frame URL for omni-10s; use with first_image

images
string<uri>[]
reference_images
string<uri>[]
videos
string<uri>[]
reference_videos
string<uri>[]
reference_video_durations
integer[]

Required with reference videos for kling-4.0; same length as videos; seconds 1–30. Billable seconds = output duration + sum of these values.

必填范围: 1 <= x <= 30
reference_audio_durations
integer[]

Required with reference audios for kling-4.0; same length as audios; seconds 5–30 (validation only, not billed).

必填范围: 5 <= x <= 30
audios
string<uri>[]
reference_audios
string<uri>[]
smart_queue
boolean
默认值:false

When true, a very low recent success rate keeps the task queued until recovery (up to about 2 hours)

响应

200 - application/json

Submitted

code
integer
示例:

20000

msg
string
data
object
最后修改于 2026年11月3日