CAD-Bench
← back
ASM-001 · Assembly & Mating · difficulty 4/5

Threaded coupling M16×1.5 (male+female pair)

sha256:33ce1d0bc09f01ea

§1Prompt verbatim

Two-piece threaded coupling. Male: Ø 16 × 30 mm long with M16×1.5 6g external thread, hex Ø 22 across-flats × 8 mm tall. Female: Ø 24 × 25 mm with M16×1.5 6H internal thread depth 22 mm, hex Ø 22. The pair must thread together a full 18 mm with no interference.

§2Ground-truth spec

shells1
watertighttrue
manifoldtrue
acceptance ε±0.02 mm
clearance[0, 0.4] 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
0.7708
Zoo Text-to-CAD
Zoo Text-to-CAD
Zoo (KittyCAD)
0.73112
OpenAI o4 (reasoning) → CadQuery
OpenAI o4 (reasoning) → CadQuery
OpenAI + CadQuery 2.4
0.6458
Adam (CADcrush)
Adam (CADcrush)
CADcrush
0.60611
Qwen3 Coder → CadQuery
Qwen3 Coder → CadQuery
Alibaba + CadQuery 2.4
0.48712
DeepSeek R1 (reasoning) → CadQuery
DeepSeek R1 (reasoning) → CadQuery
DeepSeek + CadQuery 2.4
0.47613
Claude Opus 4.7 → CadQuery
Claude Opus 4.7 → CadQuery
Anthropic + CadQuery 2.4
0.46412
Claude Opus 4.7 → OpenSCAD
Claude Opus 4.7 → OpenSCAD
Anthropic + OpenSCAD 2024.06
0.4500
Claude Sonnet 4.6 → CadQuery
Claude Sonnet 4.6 → CadQuery
Anthropic + CadQuery 2.4
0.44513
GPT-5 → CadQuery
GPT-5 → CadQuery
OpenAI + CadQuery 2.4
0.43112
Gemini 2.5 Pro → OpenSCAD
Gemini 2.5 Pro → OpenSCAD
Google + OpenSCAD 2024.06
0.3860
Gemini 2.5 Flash → CadQuery
Gemini 2.5 Flash → CadQuery
Google + CadQuery 2.4
0.37114
CAD-Coder R1
CAD-Coder R1
CAD-Coder Labs (research)
0.32814
GPT-5 Mini → OpenSCAD
GPT-5 Mini → OpenSCAD
OpenAI + OpenSCAD 2024.06
0.31117
DeepCAD
DeepCAD
Wu et al. 2021 (research)
0.26317
Llama 3.3 70B → OpenSCAD
Llama 3.3 70B → OpenSCAD
Meta + OpenSCAD 2024.06
0.23422
Claude Haiku 4.5 → CadQuery
Claude Haiku 4.5 → CadQuery
Anthropic + CadQuery 2.4
0.22224
Trellis 3D
Trellis 3D
Microsoft Research
0.0000
Spline AI
Spline AI
Spline.design
0.0000
Hunyuan3D-2
Hunyuan3D-2
Tencent
0.000104

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
AgentWatert.Manif.FeatRecMating ClearanceFit-Class ComplianceP@1p50latencycost
Human Baseline (Mech-E)0.9650.8410.8080.7641.000621.8s$5.263
Zoo Text-to-CAD0.9530.7550.6930.4950.0006.3s$0.160
OpenAI o4 (reasoning) → CadQuery0.9430.7950.6310.4530.000102.2s$0.978
Adam (CADcrush)0.9390.6460.6440.5080.00010.2s$0.226
Qwen3 Coder → CadQuery0.9270.5840.4940.2150.00019.2s$0.034
DeepSeek R1 (reasoning) → CadQuery0.9280.6260.5620.3090.000116.8s$0.041
Claude Opus 4.7 → CadQuery0.9150.7760.6360.4580.00046.7s$0.293
Claude Opus 4.7 → OpenSCAD0.9150.5940.4790.2310.00022.8s$0.313
Claude Sonnet 4.6 → CadQuery0.9190.7310.6200.3880.00013.5s$0.077
GPT-5 → CadQuery0.9190.6520.5430.3950.00036.4s$0.232
Gemini 2.5 Pro → OpenSCAD×0.9060.5050.4780.2160.00034.6s$0.087
Gemini 2.5 Flash → CadQuery×0.9040.5430.5170.2430.00011.9s$0.020
CAD-Coder R1×0.8990.6670.4720.2550.0005.6s$0.006
GPT-5 Mini → OpenSCAD×0.8960.3960.3430.1160.00010.8s$0.011
DeepCAD×0.8930.4750.3950.1670.0005.4s$0.021
Llama 3.3 70B → OpenSCAD×0.8870.4300.3990.1120.00015.8s$0.022
Claude Haiku 4.5 → CadQuery×0.8850.4970.3630.1760.00010.0s$0.024
Trellis 3D×0.8500.2180.0950.0050.00015.0s$0.043
Spline AI×0.8500.0900.0500.0010.0009.0s$0.042
Hunyuan3D-2×0.8500.2100.0820.0050.00026.1s$0.073