Seedance 2.5
ByteDance · AI videoSeedance 2.5 is ByteDance's flagship multimodal video model: it generates high-quality video with synchronized audio from text, images, video clips, and audio in one pass. Compared with Seedance 2.0, 2.5 adds native 30-second generation, larger reference sets (up to 30 images, 10 videos, and 10 audio files), stronger instruction following, and multilingual prompt support.
On VidMachine, Seedance 2.5 runs via Replicate at 480p (20 credits/s) or 720p (46 credits/s). It supports AI Video and AI Influencer projects, including lip-sync via scene audio.
Key features and benefits
Native 30-second generation
Create full 30-second clips in one pass without stitching. Duration can be set from 4 to 30 seconds, or -1 for intelligent duration when lip-syncing to scene audio.
Large multimodal reference sets
Combine up to 30 images, 10 video clips, and 10 audio files in a single generation. Reference them in prompts as [Image1], [Video1], [Audio1], etc.
Native synchronized audio
Audio and video are generated together—dialogue, effects, and music stay aligned with picture. Put spoken words in double quotes for lip-synced delivery.
Editing and extension
Provide a reference video and describe edits or continuations. The model preserves motion and style while applying your changes.
Technical specifications
Developer / familyByteDance Seedance (2.5 generation)
Resolution on VidMachine480p or 720p (project model setting)
Duration4–30 seconds per generation; -1 for audio-driven lip-sync
InputsText prompt; optional first/last frame images; optional reference images, videos, and audio
Pricing on VidMachine20 credits/s at 480p; 46 credits/s at 720p
Use cases and applications
Seedance 2.5 suits longer social clips, narrative scenes, and influencer content where you need native audio, lip-sync, and up to 30 seconds in a single generation.
Teams upgrading from Seedance 2.0 benefit from longer clips, larger reference sets, and stronger prompt adherence—with a cheaper 480p option when resolution matters less than cost.
Why this model
Choose Seedance 2.5 when you want ByteDance's latest unified audio–video model with 30-second native output, multimodal references, and a clear 480p vs 720p price tradeoff on VidMachine.
How VidMachine uses it
VidMachine routes Seedance 2.5 through Replicate at the resolution you pick in project settings (480p or 720p). For AI Influencer projects, scene audio is passed as reference audio with duration -1 so the output matches speech length.
What you should know
How does Seedance 2.5 compare to Seedance 2.0 on VidMachine?
Seedance 2.5 supports up to 30-second clips (vs 15s on 2.0) and larger reference sets. Both offer 480p and 720p: Seedance 2.5 is 20 credits/s (480p) or 46 credits/s (720p); Seedance 2.0 is 12 credits/s (480p) or 24 credits/s (720p).
Is Seedance 2.5 available for AI Influencer projects?
Yes. Seedance 2.5 supports lip-sync via scene audio, like Seedance 2.0 and WAN 2.7.