Seedance 2.5 turns one prompt into a 30-second 4K film.

ByteDance's Seedance 2.5 generates cinematic video from text or images — native 4K with 10-bit color, single-shot 30-second clips, synced native audio, and up to 50 multimodal references. No stitching, no upscaling fakery.

My Works

No videos yet

Describe a scene and generate your first clip.

Made with Seedance 2.5

A glimpse of what cinematic, controllable AI video looks like.

Fantasy Market Encounter

Fantasy Market Encounter

Server-Room Intrusion

Server-Room Intrusion

Rainy Cafe, Two Angles

Rainy Cafe, Two Angles

A video model that finally goes the distance

Seedance 2.5 pushes past the limits that held back earlier AI video — longer, sharper, and more controllable, with roughly 20% better prompt adherence than the previous generation.

Generate a full 30-second shot in a single pass — no segment stitching, with natural scene changes and tempo shifts inside one clip.

Seedance 2.5 cinematic frame of a couple on a coastal beach beneath a castle

From idea to video in three steps

No timeline, no rendering farm — just describe what you want.

1

Describe or upload

Write a prompt, or drop in reference images to guide style, character, and motion.

2

Pick your settings

Choose duration, resolution, and aspect ratio. Up to 30 seconds, native 4K.

3

Generate & download

Seedance 2.5 renders your clip in one pass. Preview, refine, and export.

A video model that finally goes the distance

Seedance 2.5 pushes past the limits that held back earlier AI video — longer, sharper, and more controllable, with roughly 20% better prompt adherence than the previous generation.

Native 30-second clips

Generate a full 30-second shot in a single pass — no segment stitching, with natural scene changes and tempo shifts inside one clip.

Native 4K, 10-bit color

Real native 4K with 10-bit color depth — smoother gradients, cleaner grading, every frame ready for a production timeline. Not 720p upscaled.

Up to 50 references

Feed in up to 50 multimodal references — images, audio, style refs, even 3D white models — a big jump from the 12 the previous version allowed.

Text & image to video

Start from a text prompt or drive the motion from reference images. Audio is generated in sync, in the same pass as the visuals.

Edit without regenerating

Tweak a finished clip while preserving its original style — refine the result instead of rolling the dice on a full regeneration.

3D previsualization

Block out shots with a 3D white-box preview to direct camera, composition, and motion before committing to the full render.

Pricing

Credits convert to video seconds. Start free, scale as you create.

Starter

$12.9

Best for getting started.

  • 100 credits
  • AI image and video generation
  • Image and video reference inputs
  • Flexible image and video formats
  • Fast cloud processing
  • Standard support

Standard

$39.9

For frequent creators and teams.

  • 400 credits
  • AI image and video generation
  • Image and video reference inputs
  • Flexible image and video formats
  • Priority processing queue
  • Priority support

Premium

$99.9

For professional creative production.

  • 1400 credits
  • AI image and video generation
  • Image and video reference inputs
  • Advanced image and video controls
  • Fastest processing speed
  • Dedicated support

Frequently asked questions

Everything you need to know before getting started.






Be first on Seedance 2.5.

Seedance 2.5 hits public launch in early July 2026. Join the early-access list now and be ready to generate cinematic 30-second 4K video on day one.