Pyyan / Video Models / Dreamina Seedance 2.5

Dreamina Seedance 2.5

Best image-to-video with audio, with 30 second clips and up to 50 references.

currentByteDance ↗Verified 2026-11-19

1156

Arena Elo

Specifications

Arena Elo1156source · as of 2026-11-19
Native audioYes, generated jointlysource · as of 2026-11-19
Max length30s, extendablesource · as of 2026-11-19
Open weightsNosource · as of 2026-11-19

What matters

  • #1 on Artificial Analysis image-to-video with audio (1161); 1156 on text-to-video with audio
  • Up to 30 seconds in one pass, extendable, with up to 50 image, video and audio references
  • Edit a specific timestamp or character in a clip without starting again
  • In Dreamina and Jimeng, and on BytePlus ModelArk; the priciest model here on AA's per-minute figure

Best for image-to-video and reference-driven ads.