Alibaba Wan 3 unifies four video modes, adds 30-second generation and native audio
- Wan 3.0 is available on Segmind as one
wan3.0-videoendpoint that infers text-to-video, image-to-video, reference-to-video, or video editing from attached inputs, replacing Wan 2.7's four separate model slugs. - The model generates up to 30-second videos at 30fps, twice Wan 2.7's 15-second limit; Segmind lists only Seedance 2.5 with the same 30-second ceiling.
- Native audio is enabled by default and generates synchronized multilingual voice and sound in the same pass; reference audio can carry voice consistency, and Segmind says the audio flag does not change pricing.
- The endpoint accepts a prompt, first and last frames, up to 10 reference images, up to five reference-video clips, and up to five reference-audio clips; video and audio references have a 15-second total limit.
- Segmind defaults to 720P and 16:9, while Alibaba documents 1080P and adaptive aspect ratio as defaults; supported resolutions are 480P, 720P, and 1080P, with duration from 2 to 30 seconds.