Home / Methodology / Test logs / t-2026-10-19-3d-evidence
Evidence-based evaluation (not tested by Loadout) · 2026-10-19 · Game Dev: 3D & Art Assets · harness n/a (third-party benchmark)

Evidence-based evaluation (not tested by Loadout): 3D Arena

3D Arena (Hugging Face, human pairwise votes on rendered outputs, 24 models, fetched 2026-10-19) lists Meshy-6 at #3 (Elo 1355, 1,019 votes), Meshy-7 at #5 (1306, 208 votes) and Meshy-5 at #23. Tripo's commercial models, Rodin and Scenario don't appear. TripoSR (#20) is an open-source research model, not the Tripo product we rank. The evidence supports Meshy's output quality in absolute terms but gives no head-to-head against our other ranked tools, so ranks are unchanged.

Rank impact: none. No score or rank was changed by this log.

Method

  1. Fetched the public leaderboard JSON behind the 3D Arena Space (vote_type=render).
  2. Matched entries to ranked tools by name only. Unmatched tools are reported as absent.

Results

toolentries
MeshyMeshy-6 #3 (1355, 1019 votes); Meshy-7 #5 (1306, 208 votes); Meshy-5 #23 (1001, 958 votes)
TripoNo entry for Tripo's commercial models (TripoSR open-source research model at #20 is not the ranked product)
Rodin (Hyper3D)No entry
ScenarioNo entry

Not covered

  • tripo
  • rodin
  • scenario (absent from leaderboard)

Limitations

Human-preference votes on renders. Doesn't measure topology, UVs, rigging or engine import (our Q and E criteria). Small vote count for Meshy-7.

Sources

Raw files