# On the Opportunities and Risks of Frontier Models for 3D Modeling, Computational Design and Robotics

A continuously updated survey that collects and verifies public demonstrations of frontier multimodal models doing 3D modeling, computational design and robotics work, assessing what they can reliably do today.

This repository maintains a live, continually updated survey of what frontier multimodal AI models can do in 3D modeling, computational design and robotics. It collects and verifies 328 demonstrations and their accompanying benchmark evaluations across 232 case groups.

The survey's findings cover three areas: models as drafting tools that produce editable CAD scenes and parametric assemblies, robotics applications where models are most effective for offline controller synthesis rather than fast closed-loop control, and how the software interface between a model and its tools affects reported performance.

The project is maintained by the Computational Design and Fabrication Group at MIT CSAIL and is presented as an interactive web portal alongside the repository.

Website: https://mit-cdfg.github.io/Survey-AI-for-3D-modeling-Robotics/