REVENG-002 · Reverse Engineering · difficulty 4/5
Three-view ortho → bracket
sha256:30d72b41ae9c0014…
§1Prompt verbatim
From the supplied 1:1 front/top/side dimensioned drawing (PNG, 600 dpi) reproduce the part. All dimensions and tolerances on the drawing are authoritative.
§2Ground-truth spec
shells1
watertighttrue
manifoldtrue
acceptance ε±0.1 mm
§3Reference render
canonical reference · drag to orbit, scroll to zoom
Visualisation is rebuilt in-browser from the canonical parametric description. Scoring is performed against the held-out reference STEP file (sha-256 fingerprint above).
§4Per-agent renders
reference + 10 agent outputs · scored against the held-out STEP
vol IoU · BREP · manifold
canonical reference
REFERENCE
canonical · ground truth
1.000100✓
Human Baseline (Mech-E)
Human Baseline (Mech-E)
n=4 senior engineers
0.6548✗
Hunyuan3D-2
Hunyuan3D-2
Tencent
0.57311✗
Claude Sonnet 4.6 → CadQuery
Claude Sonnet 4.6 → CadQuery
Anthropic + CadQuery 2.4
0.5489✗
GPT-5 → CadQuery
GPT-5 → CadQuery
OpenAI + CadQuery 2.4
0.52510✗
Trellis 3D
Trellis 3D
Microsoft Research
0.5210✗
OpenAI o4 (reasoning) → CadQuery
OpenAI o4 (reasoning) → CadQuery
OpenAI + CadQuery 2.4
0.51610✗
Gemini 2.5 Pro → OpenSCAD
Gemini 2.5 Pro → OpenSCAD
Google + OpenSCAD 2024.06
0.4690✗
Gemini 2.5 Flash → CadQuery
Gemini 2.5 Flash → CadQuery
Google + CadQuery 2.4
0.43714✗
Claude Opus 4.7 → CadQuery
Claude Opus 4.7 → CadQuery
Anthropic + CadQuery 2.4
0.41613✗
Claude Opus 4.7 → OpenSCAD
Claude Opus 4.7 → OpenSCAD
Anthropic + OpenSCAD 2024.06
0.4110✗
Zoo Text-to-CAD
Zoo Text-to-CAD
Zoo (KittyCAD)
0.39215✗
DeepSeek R1 (reasoning) → CadQuery
DeepSeek R1 (reasoning) → CadQuery
DeepSeek + CadQuery 2.4
0.37915✗
CAD-Coder R1
CAD-Coder R1
CAD-Coder Labs (research)
0.35917✗
GPT-5 Mini → OpenSCAD
GPT-5 Mini → OpenSCAD
OpenAI + OpenSCAD 2024.06
0.34315✗
Claude Haiku 4.5 → CadQuery
Claude Haiku 4.5 → CadQuery
Anthropic + CadQuery 2.4
0.33418✗
Qwen3 Coder → CadQuery
Qwen3 Coder → CadQuery
Alibaba + CadQuery 2.4
0.33018✗
Adam (CADcrush)
Adam (CADcrush)
CADcrush
0.30619✗
Llama 3.3 70B → OpenSCAD
Llama 3.3 70B → OpenSCAD
Meta + OpenSCAD 2024.06
0.23520✗
Spline AI
Spline AI
Spline.design
0.2240✗
DeepCAD
DeepCAD
Wu et al. 2021 (research)
0.09052✗
Each tile is rebuilt from the canonical parametric description and degraded to match the agent's scored profile (tessellation, non-manifold face removal, dimension scale jitter, missing features). Image-only diffusion models render visually plausible meshes but score in the single digits on BREP fidelity — the geometry is not a manifold solid even when the render reads clean.
§5Per-agent metrics
ranked by Vol IoU · same data as the leaderboard, restricted to this task
| Agent | Vol IoU | Watert. | Manif. | Named-Dimension RMSE | FeatRec | P@1 | p50 | latency | cost |
|---|---|---|---|---|---|---|---|---|---|
| Human Baseline (Mech-E) | 0.654 | ✓ | 0.945 | 0.145 | 0.956 | 0.000 | — | 806.4s | $5.561 |
| Hunyuan3D-2 | 0.573 | ✓ | 0.932 | 0.551 | 0.212 | 0.000 | — | 25.4s | $0.078 |
| Claude Sonnet 4.6 → CadQuery | 0.548 | ✓ | 0.933 | 0.226 | 0.711 | 0.000 | — | 15.5s | $0.075 |
| GPT-5 → CadQuery | 0.525 | ✓ | 0.933 | 0.213 | 0.716 | 0.000 | — | 40.9s | $0.173 |
| Trellis 3D | 0.521 | ✓ | 0.933 | 0.557 | 0.204 | 0.000 | — | 15.6s | $0.047 |
| OpenAI o4 (reasoning) → CadQuery | 0.516 | ✓ | 0.933 | 0.181 | 0.783 | 0.000 | — | 98.7s | $1.293 |
| Gemini 2.5 Pro → OpenSCAD | 0.469 | ✓ | 0.923 | 0.221 | 0.524 | 0.000 | — | 30.1s | $0.080 |
| Gemini 2.5 Flash → CadQuery | 0.437 | ✓ | 0.919 | 0.270 | 0.577 | 0.000 | — | 13.3s | $0.023 |
| Claude Opus 4.7 → CadQuery | 0.416 | ✓ | 0.916 | 0.205 | 0.754 | 0.000 | — | 27.2s | $0.337 |
| Claude Opus 4.7 → OpenSCAD | 0.411 | ✓ | 0.917 | 0.253 | 0.533 | 0.000 | — | 33.6s | $0.266 |
| Zoo Text-to-CAD | 0.392 | × | 0.906 | 0.201 | 0.738 | 0.000 | — | 5.6s | $0.188 |
| DeepSeek R1 (reasoning) → CadQuery | 0.379 | × | 0.908 | 0.261 | 0.613 | 0.000 | — | 85.8s | $0.044 |
| CAD-Coder R1 | 0.359 | × | 0.903 | 0.293 | 0.650 | 0.000 | — | 4.5s | $0.004 |
| GPT-5 Mini → OpenSCAD | 0.343 | × | 0.903 | 0.313 | 0.386 | 0.000 | — | 10.0s | $0.009 |
| Claude Haiku 4.5 → CadQuery | 0.334 | × | 0.900 | 0.340 | 0.536 | 0.000 | — | 5.8s | $0.016 |
| Qwen3 Coder → CadQuery | 0.330 | × | 0.902 | 0.303 | 0.629 | 0.000 | — | 16.9s | $0.024 |
| Adam (CADcrush) | 0.306 | × | 0.900 | 0.170 | 0.730 | 0.000 | — | 7.0s | $0.266 |
| Llama 3.3 70B → OpenSCAD | 0.235 | × | 0.886 | 0.322 | 0.425 | 0.000 | — | 18.6s | $0.022 |
| Spline AI | 0.224 | × | 0.886 | 0.556 | 0.100 | 0.000 | — | 7.1s | $0.048 |
| DeepCAD | 0.090 | × | 0.865 | 0.333 | 0.435 | 0.000 | — | 4.3s | $0.021 |