Four-Factor Integrity Breakdown & Primary-Source Evidence
Privacy
Weight 40%
5.5/10
Access Model: Public Prompts / Private Answers
"Quality Elo comes entirely from blind, pairwise preference votes cast by our recruited panel. In each comparison, a panelist sees two scenes built for the same task by two different models, interacts with both, and selects the one they prefer. Public votes cast in AA-3D Voting do not enter the rating calculation."
"Create an interactive adult man performing weightlifting exercises, with views that expose his muscles, skeleton and organs in anatomically accurate form."
Originality
Weight 25%
10/10
Reuse Profile: All Original (0% reuse)
"AA-3D v1.0 measures how well AI models build interactive 3D experiences. Each model receives a written brief and builds a single web page with three.js in an offline sandbox."
Freshness
Weight 20%
8/10
Initial Release: 2026-10Latest Refresh: 2026-10Rolling Updates: No
"Elo from blind pairwise votes · 95% confidence intervals · Higher is better · Data as at 12 October 2026"
Integrity: Independent Offline (Leak-Verified)Sandbox: offline isolated
"We run every model in Stirrup, our open-source agent harness. Each task runs in its own sandbox with no network access."
Frontier Model Standings on AA-3D
Top 15 active models tracked
Benchmark maintainers can embed this live SVG badge in GitHub READMEs or documentation:
[](https://benchaudit.org/benchmarks/aa-3d/)