Single 30-Second Clips Expanded to Up to 3 Minutes...ByteDance Opens the Era of Long-Form AI Video with 'Seedance 2.5'

ByteDance has unveiled Seedance 2.5, an AI video model built for longer storytelling and multimodal references.

It can generate up to 30 seconds of audio-video at a time, and multiple segments can be stitched together into videos lasting minutes.

Dreamina AI also offers a long-form mode that can produce videos up to 3 minutes, expanding beyond short-form generation.

The model supports image, video, and audio references, plus framing, camera, lighting, and background editing for more professional workflows.

The release signals AI video moving beyond clips into long-form content, with implications for education, industry, and synthetic data creation.