Seedance 2.5 Flash — Multimodal Image & Text-to-Video
What is Seedance 2.5 Flash?
Seedance 2.5 Flash is the fast, economical tier of ByteDance Seed's Seedance 2.5, a multimodal audio-video generation model. Create 4–30 second clips at 480p or 720p from a text prompt, a first or last frame, or reference images, videos and audio — with synchronized native audio rendered in the same pass as the picture. The Flash tier trades the highest resolutions for faster turnaround and lower cost than Seedance 2.5, making it the tier for iterating and producing video at volume.
Key Features
- Native synchronized audio — dialogue, sound effects, ambience and music are generated together with the video and land in sync with the on-screen motion, not dubbed on afterwards.
- Multimodal referencing — up to 10 reference images, 5 reference videos and 5 reference audio clips, cited inline as @Image, @Video and @Audio, to keep characters, style and motion consistent.
- First- and last-frame control and multi-shot "Shot 1:" scripting for sequenced scenes.
- Video editing and extension with timestamp-level control — adjust or lengthen a clip in place instead of regenerating it.
- Flexible output — 4–30 second durations (or auto), 480p/720p, selectable aspect ratios, and an optional returned last frame for chaining shots.
Best Use Cases
Seedance 2.5 Flash suits social and short-form video, product and explainer clips, storyboard animatics, dialogue scenes that need lip-synced sound, and iterative pipelines where many 480p drafts get promoted to 720p finals. A reference text-to-video render returned a clean 1280×720 clip at 24fps with synchronized stereo audio.
Prompt Tips and Output Quality
Write concrete subject, action, camera move and the sound you want to hear, and keep generate_audio on for any diegetic scene. Use a fixed aspect ratio for text-to-video and adaptive when supplying a first frame or editing a clip. Fix the seed to iterate the same scene. Output is H.264 MP4 at 24fps.




