Video
Seedance 2.5
ByteDance's latest multimodal video model - up to 30-second clips, massively expanded omni-reference, and synced audio.
About this model
Seedance 2.5 (doubao-seedance-2.5) is ByteDance Seed's newest short-form video model, pushing duration to 30 seconds and dramatically expanding multimodal input. Its omni-reference mode now accepts up to 30 images, 10 videos, and 10 audio files - 50 media items total - so you can combine character look, motion style, voice, and music in a single render. Native audio-video co-generation produces synced dialogue and sound, while the same text-to-video and first/last-frame modes carry over from Seedance 2.0. Output tops out at 720p, making 2.5 the choice for longer, reference-rich storytelling rather than maximum-resolution finals. On HiArt it is the default video model: start a still in any image model, then animate it in Seedance 2.5 on the same canvas.
Best for
- Up to 30-second clips - twice the duration of Seedance 2.0
- Expanded omni-reference: 30 images, 10 videos, 10 audio (50 total)
- Native audio-video co-generation with lip sync
- Text-to-video, first/last frame, and omni-reference modes
- 480p and 720p output at up to 30 seconds
- Default video model on HiArt - animate canvas stills directly
Common workflows
- Longer short-form ads and 20-30 second storytelling clips
- Music videos and lip-synced content using audio references
- Character-driven scenes combining many reference images and clips
- Multi-shot narratives that need consistent identity and sound
- Product showcase videos with synced voiceover and sound design
How to use on HiArt
- 01Add a video node on the HiArt canvas and select Seedance 2.5.
- 02Pick a mode: text-to-video, first/last frame, or omni-reference.
- 03Attach reference media (up to 30 images, 10 videos, 10 audio) and cite them in the prompt.
- 04Enable audio, set duration (up to 30s) and resolution (480p/720p), then generate and iterate.
Capabilities
- Provider
- ByteDance Seed (APIMart doubao-seedance-2.5)
- Modes
- Text-to-video, first/last frame, omni-reference
- Duration
- 4-30 seconds
- Resolution
- 480p, 720p
- Omni-reference limits
- 30 images, 10 videos, 10 audio (50 total)
- Audio
- Native generation with lip sync
Learn more
- What Is Seedance 2.5? — HiArt Blog
Product overview covering 30-second duration, expanded omni-reference, modes, and when to choose 2.5 vs 2.0.
- Seedance 2.5 Prompt Guide — HiArt Blog
@ reference syntax, 30-second timing beats, camera language, and copy-paste templates for longer clips.
- Doubao Seedance 2.5 - APIMart docs
Official API reference for Seedance 2.5 including modes, omni-reference limits, and prompting.
- Seedance - ByteDance Seed
Official Seedance model family overview from ByteDance Seed.
FAQ
- How is Seedance 2.5 different from Seedance 2.0?
- Seedance 2.5 doubles max duration to 30 seconds and massively expands omni-reference - up to 30 images, 10 videos, and 10 audio files versus 2.0's 9 images, 3 videos, and 3 audio. The tradeoff is resolution: 2.5 tops out at 720p while 2.0 reaches 1080p. Use 2.5 for longer, reference-rich clips; use 2.0 for 1080p finals.
- Does Seedance 2.5 generate audio?
- Yes. Native audio-video co-generation produces synced dialogue, ambient sound, and music, with lip sync support.
- What is the max video length?
- Up to 30 seconds at 480p or 720p - the longest of any Seedance model on HiArt.
Related models
Try on the canvas
Sign up free, open a project, and set model preferences before you generate.