ByteDance Releases Seedance 2.5 Video Model

Dreamina has launched Seedance 2.5, its latest AI video generation model. Sources describe native 30-second output, a long video mode reaching three minutes in one shot, and added audio support. The model also includes interactive frame editing and up to 50 multimodal references for greater control. Availability is rolling out regionally, with confirmed access in parts of Europe but not yet for all users. Commentary highlights the length increase as a notable step beyond prior video generation limits, positioning the release as focused on cinematic and storytelling use cases.