gnucleus-ai/cad-bench

Coding

Professional

gNucleus AI CAD-generation benchmark v3 – 100 parametric FreeCAD tasks: 30 text-to-CAD create, 30 create-and-edit, 40 image-to-CAD (Harbor schema 1.3; requires harbor >= 0.20.0).

← Back to Registry

Run this task

CLI:

inspect eval inspect_harbor/gnucleus_ai_cad_bench --model openai/gpt-5

Python:

from inspect_ai import eval
from inspect_harbor import gnucleus_ai_cad_bench

eval(gnucleus_ai_cad_bench(), model="openai/gpt-5")

Dataset information

Harbor registry	gnucleus-ai/cad-bench
Inspect task	`gnucleus_ai_cad_bench`
Latest digest	sha256:8a786725d3648a76282fc5868df06da3d5e00d603f283573a53430803465d78a
Samples	100
Source	https://www.gnucleus.ai/cad-bench

See Task Parameters for the parameter set shared across all Harbor tasks.