Top3D.AI Arena / Final Results

Season 5

March 26 / Mar 1, 2026 – Apr 1, 2026

Season Cup / 1 of 4 arena crowns / 4 podium places

20,437 blind votes / 22 tools / 4 arenas

The story of the season

  • The Season Cup changed hands: Tripo AI took it from Hitem3D (Sparc3D) with 1 of 4 arena crowns.
  • The climb of the season: Hunyuan 3D Retopo, +211 ELO in Low Poly vs last season's final.
  • The strongest first showing: YVO3D 2.5 entered Geometry at #1 of 22.

Full standings

Textured22 competed / 13,655 votes
Geometry22 competed / 3,242 votes
Low Poly6 competed / 2,489 votes
Segmentation4 competed / 1,051 votes

What changed on top3d.ai this season

How champions are decided

Championship rules v1 / in effect since Season 1

Four arenas, four crowns

  • Textured, Geometry, Low Poly, and Segmentation are four separate competitions with separate votes and separate results. Textured and Geometry share the same model roster and the same prompts; nothing else.
  • Every model starts the season at 1,000 ELO. Ratings come only from this season's blind votes, with no branding shown before a vote.
  • The crown goes to the highest final ELO when the season closes. Exact ties break by season votes, then alphabetically.

The Season Cup

  • The Cup is the season's headline title. It goes to the vendor whose models take the most arena crowns.
  • Winning it takes at least one Main Arena crown (Textured or Geometry). Ties break like a medal table: most crowns, then most 2nd places, then most 3rd places. A still-tied Cup goes to the vendor whose best model finished the season with the higher ELO.
  • Closed results are never restated, and rankings are not for sale. Every record on this page follows the same published standings.
Season 6 is live
Every vote reshapes the next finale.
Blind battles, real models, no branding until you vote.
Share the finale

“Season 5 (March 26) of the top3d.ai Arena is complete: Tripo AI takes #1. 20k+ community votes collected.” / https://www.top3d.ai/leaderboard/seasons/mar-26

Permanent result / Open methodology