bytedance/seedance-2.5 is ByteDance's advanced multimodal AI video generation model, designed for long-form storytelling, video editing, extension, and reference-driven video creation. It can combine text with multiple visual and audio references to create coherent videos with strong control over characters, scenes, and cinematic direction. Seedance 2.5 supports up to 50 image, video, and audio reference assets, along with first-frame and first-and-last-frame control. It also supports optional generated audio and multilingual audiovisual generation, making it suitable for sophisticated creative and production workflows.