Skip to content

Seedance 2.5

ByteDance's video and audio model, launched on July 31, 2026, with single-pass clips of up to 30 seconds and many image, video and audio references. It is in the second group on LMArena text-to-video and third on Artificial Analysis' text-to-video arena.

ByteDance SeedVideo modelReleased July 2026Data checked October 18, 2026

This model is not in the current people's rankings: it is either new and still collecting votes, or ranked below the top 12. See the AI leaderboard for the models people rate highest.

Strengths and weaknesses

  • Third on Artificial Analysis' text-to-video arena
  • Fourth-highest score on LMArena image-to-video
  • Single-pass clips of up to 30 seconds
  • Slightly below the older Seedance 2.0 on LMArena text-to-video (1473.9 vs 1479.2)
  • BytePlus, its API platform, is not available in the United States, according to BytePlus

Best for

  • Longer clips with consistent characters
  • Videos built from many reference images

Every score we have

Each bar compares this model with the best result any model on this page has on the same test.

  • LMArena text to video1,473.9Arena score 1473.9 (95% CI 1465.4 to 1482.5); 8,001 votes; entry "dreamina-seedance-2.5-720p"; position 7 of 48 on the full board
  • LMArena image to video1,477.4Arena score 1477.4 (95% CI 1470.2 to 1484.6); 15,729 votes; entry "dreamina-seedance-2.5-720p"; position 4 of 48 on the full board
  • Artificial Analysis video arena1,160.1Elo 1160.0 (95% CI 1145.0 to 1175.0); 5,939 appearances; position 3 of 32 on the Artificial Analysis text-to-video arena
  • Price per minute of video (estimate)$34$34.12 per minute of video (Artificial Analysis estimate)

Compare with