SHEET-003 · Sheet-Metal Bodies · difficulty 3/5

U-channel bracket, 1.5 mm Al, k=0.40

sha256:311fac0bb472ea91…

§1Prompt verbatim

Sheet-metal U-channel from 1.5 mm Al-5052: outside dimensions 80 × 30 × 25 mm tall, bend radius 1.5 mm (inside), k-factor 0.40. Two M5 clearance holes (Ø 5.5) on the base, centred 15 mm from each end. Output the folded body and report the unfolded flat pattern dimensions.

§2Ground-truth spec

shells1

watertighttrue

manifoldtrue

acceptance ε±0.1 mm

featuresbend_x2, M5_clearance_x2

§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)

n=4 senior engineers

0.8749✓

Claude Opus 4.7 → CadQuery

Anthropic + CadQuery 2.4

OpenAI + CadQuery 2.4

0.55614✗

DeepCAD

Wu et al. 2021 (research)

0.43414✗

Claude Opus 4.7 → OpenSCAD

Anthropic + OpenSCAD 2024.06

0.3740✗

Gemini 2.5 Pro → OpenSCAD

Google + OpenSCAD 2024.06

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.	Named-Dimension RMSE	FeatRec	Wall-Thickness Uniformity	P@1	p50	latency	cost
Human Baseline (Mech-E)	✓	0.979	0.150	0.823	0.822	1.000	—	578.4s	$4.900
Claude Opus 4.7 → CadQuery	✓	0.945	0.264	0.685	0.683	0.000	—	45.4s	$0.307
Adam (CADcrush)	✓	0.940	0.208	0.655	0.678	0.000	—	7.2s	$0.269
Zoo Text-to-CAD	✓	0.942	0.141	0.730	0.706	0.000	—	7.6s	$0.201
GPT-5 → CadQuery	✓	0.929	0.234	0.736	0.633	0.000	—	43.1s	$0.234
DeepCAD	✓	0.914	0.275	0.497	0.356	0.000	—	4.2s	$0.019
Claude Opus 4.7 → OpenSCAD	×	0.908	0.289	0.575	0.513	0.000	—	25.8s	$0.304
Gemini 2.5 Pro → OpenSCAD	×	0.908	0.242	0.486	0.512	0.000	—	33.1s	$0.101
Trellis 3D	×	0.881	0.544	0.218	0.128	0.000	—	14.7s	$0.057
Spline AI	×	0.857	0.568	0.085	0.061	0.000	—	7.2s	$0.045