Seedance 2.5 is a next-generation multimodal AI video model built for videos up to 30 seconds, native 4K model output, richer reference control, precise local editing, and production-ready creative workflows. On Wan 3.0, choose strict first and last frames or combine image, video, and audio references to generate 4–30-second or smart-duration video in 480p, 720p, or 1080p.
Bring a character to life with a performance you choose. Upload a character image and a motion video, then shape the animation with orientation, background, and 720p or 1080p output controls.
Turn text or a reference image into a video draft in one focused workflow. Compare motion, prompt-guided direction, character references, and synchronized native audio while keeping the cost of each creative test visible.
Generate or refine reference visuals that can guide product videos, character looks, scenes, and style direction.
Reference image creation
Product and scene visuals
Style direction for video
Powered by Seedance 2.5Seedance 2.5
Create video
Describe your video
0/30000
Duration · Estimated credits
5 seconds·195 credits
4 seconds30 seconds
My videosWorks
Model overview
What is Seedance 2.5?
Seedance 2.5 is designed for longer video generation, native 4K model output, multimodal reference control, precise local editing, and more mature production workflows.
Videos up to 30 seconds
Native 4K model output
Up to 50 multimodal references
3D blockout-assisted planning
Product Features
Seedance 2.5 creative capabilities
Seedance 2.5 adds videos up to 30 seconds, native 4K model capability, as many as 50 multimodal references, 3D blockout-assisted planning, and more professional production control. The current Wan 3.0 workspace supports output up to 1080p.
01
Videos up to 30 seconds
Create a longer scene in one generation, with more room for story development, continuous action, and deliberate pacing.
02
Native 4K model output
Seedance 2.5 is designed for native 4K output with sharper detail and more post-production latitude; the current Wan 3.0 workspace supports up to 1080p.
03
Up to 50 multimodal references
Combine as many as 30 images, 10 videos, and 10 audio files with text direction to keep characters, products, motion, style, and sound aligned.
04
3D blockout-assisted planning
Plan scene layout, camera paths, and movement with a simple 3D blockout before committing to the final generation.
05
Licensed character and brand workflows
Organize approved characters, stories, and brand assets into one reference-led brief while keeping rights and release review in the production process.
Model selection guide
Seedance 2.5 workflows
Seedance 2.5 advances video length, native 4K model output, multimodal reference control, and pre-production planning, making it better suited to longer scenes, brand assets, and professional video production.
Feature
Seedance 2.5
Seedance 2.0
Model focus
Next-generation production model
Classic multimodal video model
Maximum video length
Up to 30 seconds
Shorter-video workflow
Native 4K output
Supported
Standard output
Multimodal references
Up to 50 assets
Basic multimodal references
Pre-production planning
3D blockout assistance
Basic shot planning
Local editing and control
More precise and flexible
Standard control
Best for
Long-form stories, brands, professional production
Everyday video creation
Production workflow
How to use Seedance 2.5
Move through shot direction, reference preparation, output settings, and result review in a clear sequence.
Select Seedance 2.5
Advantages
Smarter reference control with Seedance 2.5
Keep the creative intent aligned through generation and revision with stronger multimodal understanding, reference-led editing, and production controls.
Precise multimodal understanding
Start a new project
Put Seedance 2.5 into your next video workflow
Return to the workspace to shape a prompt, prepare references, compare a direction, and start the next generation task.
Seedance 2.5 is a multimodal AI video workflow available on Wan 3.0 for text, strict first/last-frame, and image, video, or audio reference generation with 480p, 720p, or 1080p output.
How many references can Seedance 2.5 use?
Multimodal mode accepts up to 30 images, 10 videos, and 10 audio files. Strict first/last-frame mode is separate and cannot be combined with those reference arrays.
What can creators build with Seedance 2.5?
Typical projects include product launches, ecommerce videos, branded stories, virtual-influencer content, short films, trailers, cinematic concepts, and reference-led campaign assets.
Which Seedance 2.5 reference workflow should I use?
Use strict first-frame or first-plus-last-frame guidance when endpoints must be anchored. Use multimodal mode for image, video, and audio reference arrays. These source modes cannot be mixed.
What duration does Seedance 2.5 support?
Choose 4-30 seconds or smart duration. The workspace initially selects 5 seconds.
What Seedance 2.5 resolutions are available?
Available resolutions are 480p, 720p, and 1080p. The workspace initially selects 720p, while 1080p is the highest supported tier.
Can I try Seedance 2.5 with credits?
Yes. Without video input, Seedance 2.5 costs 18, 39, or 71 credits per output second at 480p, 720p, or 1080p. With reference video, it costs 11, 24, or 43 credits per second and bills input duration plus output duration. These rates currently apply with generated audio off.
