Research
The Best Computer-Use Agent Scores 17.5% on Real CAD Work Against an 87% Expert Reference
CADWorld (arXiv 2609.16251, submitted 14 Sep 2026) is a 200-task FreeCAD benchmark spanning 11 mechanical-CAD workflow categories including sketching, part modeling, assembly, CAM, FEM, and technical drawing, with agents operating through screenshots and GUI actions and success determined by executable checks over the saved native project. Across seven current agents the strongest reached 17.5% success versus an 87.0% expert reference pass. The failure profile is informative: weaker agents fail before producing a valid artifact at all, while stronger ones increasingly fail on structural, geometric, and construction-process requirements.
↳ Follow the thread