2026 Lu OmniMech Mechanical CAD Benchmark

Recent vision-language models (VLMs) have demonstrated a remarkable ability to generate executable CAD programs from images by aligning visual geometry with symbolic program representations. However, existing approaches primarily focus on coarse, parameter-free 3D object datasets, such as tables and chairs, while overlooking the fine-grained geometric and millimeter-level tolerance requirements of high-fidelity 3D CAD, whose functionality and manufacturability depend on precise dimensions. To bridge this gap, we introduce OmniMech, the first million-grade benchmark for evaluating VLMs on generating executable CAD programs from industrial mechanical manufacturing. OmniMech comprises over 251,000 fully dimensioned and toleranced industrial 2D orthographic mechanical drawings, each paired with its corresponding parametric CAD model, multi-view renderings, 3D geometric representations (mesh, STEP and B-rep), and rich semantic annotations. The benchmark comprises four tasks: (1) parametric CAD program synthesis, benchmarking VLMs against executable CAD-specialized CAD generation models that construct viable 3D objects from 2D engineering drawings; (2) diagram-to-3D reasoning, evaluating VLMs alongside CAD-specialized models on reconstructing geometry that is visually and structurally consistent with the multi-view perspective in the input drawing; (3) annotation-grounded geometric reasoning, interpreting dimensions, symbols, and feature calls and enforcing them in geometric constraints in the generated CAD program; and (4) spherical visual tool-augmented agent reasoning, selecting and invoking visualization, measurement, CAD execution, and verification tools to iteratively accomplish tasks (1)-(3). Our results reveal that VLMs and even existing state-of-the-art vision exhibit substantial limitations in industrial CAD generation, including unreliable synthesis of executable parametric programs, brittle reconstruction of fine-grained 3D geometry, and poor generative performance.