Seedance 2.5

ByteDance / Volcano Engine · video generation

Text or an image in, video out. Give it a description (or a still plus a description) and it produces a clip; the vendor advertises narratives up to 30 seconds.

What it is good for

  • Short clips that need no continuous narrative — social posts, product demos, B-roll
  • Animating a still you already have

How it bills

Billed per video token; the token count depends on duration, resolution and whether a reference video was supplied. We do not convert that into a per-second rate — that needs a formula we do not have. The vendor's own online-inference rate at 480p/720p output: ¥70 per million tokens without a reference video, ¥42 per million tokens with one. Below are the vendor's own worked examples.

A 5-second 480p clip (16:9, no reference video input)about ¥3.36 per clip (about ¥0.67/second)
A 5-second 720p clip (16:9, no reference video input)about ¥7.56 per clip (about ¥1.51/second)
A 5-second 480p clip (16:9, with a 2-30 second reference video input)¥3.63–14.12 per clip, rising with the reference video's length — the low end is a 2–4s reference, the high end a 30s reference
A 5-second 720p clip (16:9, with a 2-30 second reference video input)¥8.16–31.75 per clip, rising with the reference video's length — the low end is a 2–4s reference, the high end a 30s reference

Caveats

  • Billed in video tokens, not per second or per clip — the vendor's own price examples show 720p (about ¥7.56) costs more than twice as much as 480p (about ¥3.36) for the same 5 seconds
  • Mainland China and international access are separate channels (Volcano Engine / BytePlus ModelArk) with different onboarding and possibly different rates

Using it without writing code

If you do not write code, use it through Dreamina, Doubao or CapCut rather than the API. Open a Volcano Engine console account only when wiring it into your own pipeline.

The vendor page these figures came from

Basis: vendor claim (not third-party verified) · verified 2026-08-09