P3D-Bench - Text-to-3D Descriptive: leaderboard
P3D-Bench (2026) tests parametric 3D generation through code. On 400 Text-to-3D cases from descriptive specifications, a model writes a JSON or OpenSCAD CAD program and an MLLM judge (Gemini 3.1 Pro) answers specification-derived questions about the rendered result and rates its semantics. Programs are compiled, meshed and aligned to the ground truth, invalid outputs take the worst value, and each general model runs with its maximum thinking budget; the score is on a 0-100 scale averaged over output formats, higher is better.
Metric: MLLM-judge bucket score (0-100) (QA on the descriptive specification and semantic judging of renders) on 400 Text-to-3D cases from descriptive specifications (minimal JSON and OpenSCAD outputs), the model writes a parametric CAD program that is compiled, meshed and aligned to the ground truth; invalid outputs take the worst value; each general model runs with its maximum thinking budget; averaged over the output formats; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (xHigh) | 87.5 | #49 (GPT-5.5) |
| 2 | Gemini 3.1 Pro (Preview) (High) | 86.5 | #94 (Gemini 3.1 Pro (Preview)) |
| 3 | Claude Opus 4.6 (Max) | 85 | #102 (Claude Opus 4.6) |
| 4 | Seed 2.0 Pro (High) | 72.3 | #211 (Seed 2.0 Pro) |
| 5 | DeepSeek V4 Pro (Max) | 69.9 | #188 (DeepSeek V4 Pro) |
| 6 | Qwen 3.6 Plus (Thinking) | 66.7 | #222 (Qwen 3.6 Plus) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=p3d-bench-text-to-3d-descriptive · How It Works · Data refreshed daily, snapshot 2026-11-13.