3D-DefectBench - Geometry (Silver Labels): leaderboard
3D-DefectBench measures how well vision-language models spot defects in generated 3D assets from rendered multi-view images, marking each defect type present or absent. This board covers the five geometry defects, scored against crowd majority labels on the held-out asset split. Scores are the Matthews correlation coefficient, macro-averaged over defect types, so zero means chance agreement and one means perfect agreement.
Metric: Macro MCC (Matthews correlation from -1 to 1, macro-averaged over the five geometry defect types) of the model's binary per-defect verdicts against crowd majority labels on the 549-asset silver holdout; the fixed c004 design: six-view oblique RGB turntable renders of each text-to-3D asset and a rubric-guided checklist prompt. Source: arxiv.org. Saturation forecast: Around 2029. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 0.24 | #123 |
| 2 | Gemini 3.1 Flash Lite | 0.23 | #364 |
| 3 | GPT-5 Mini | 0.21 | #348 |
| 4 | Gemini 2.5 Pro | 0.21 | #266 |
| 5 | Claude Opus 4.7 | 0.21 | #79 |
| 6 | Gemini 3.1 Pro (Preview) | 0.21 | #94 |
| 7 | Claude Sonnet 4.6 | 0.18 | #161 |
| 8 | GPT-4o | 0.17 | #741 |
| 9 | Qwen 3.5 397B A17B | 0.15 | #246 |
| 10 | Claude Haiku 4.5 | 0.12 | #568 |
| 11 | Mistral Small 3.1 | 0.09 | #1689 |
| 12 | Qwen 2.5 VL 7B | 0.03 | #1787 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=3d-defectbench-geometry-silver-labels · How It Works · Data refreshed daily, snapshot 2026-11-13.