STD-002 · Standards Compliance · difficulty 4/5

ISO 4762 M8×30 socket-head cap screw

sha256:7af0ce123b984091…

§1Prompt verbatim

Model an ISO 4762 M8 × 30 mm socket-head cap screw, property class 12.9, with a fully-formed thread (ISO 261 6g) and a hexagonal socket sized to take a 6 mm hex key. Head Ø 13, head height 8, fillet under-head R 0.4 mm.

§2Ground-truth spec

shells1

watertighttrue

manifoldtrue

acceptance ε±0.05 mm

featuresM8x1.25_thread, hex_socket_6mm, underhead_R0.4

§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)

OpenAI + CadQuery 2.4

0.57911✗

Claude Opus 4.7 → CadQuery

Anthropic + CadQuery 2.4

Claude Opus 4.7 → OpenSCAD

Anthropic + OpenSCAD 2024.06

0.4990✗

Gemini 2.5 Pro → OpenSCAD

Google + OpenSCAD 2024.06

0.4150✗

DeepCAD

Wu et al. 2021 (research)

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	Watert.	Manif.	GD&T Compliance	Standards Compliance	p50	latency	cost
Human Baseline (Mech-E)	✓	0.952	0.900	0.944	—	518.4s	$4.932
Adam (CADcrush)	✓	0.947	0.647	0.721	—	9.7s	$0.247
GPT-5 → CadQuery	✓	0.936	0.557	0.686	—	51.2s	$0.183
Claude Opus 4.7 → CadQuery	✓	0.937	0.669	0.720	—	41.4s	$0.285
Zoo Text-to-CAD	✓	0.932	0.642	0.752	—	5.3s	$0.196
Claude Opus 4.7 → OpenSCAD	✓	0.923	0.454	0.578	—	37.5s	$0.268
Gemini 2.5 Pro → OpenSCAD	×	0.911	0.452	0.505	—	28.3s	$0.089
DeepCAD	×	0.881	0.362	0.262	—	4.5s	$0.024
Trellis 3D	×	0.850	0.054	0.047	—	10.4s	$0.049
Spline AI	×	0.850	0.027	0.019	—	9.6s	$0.045