I have been trying to force LLMs to work with geometries for over a month and it's so hard. Even the best LLMs have an extremely poor sense of geometric relationships in my testing. I would also stay away from mesh based CAD like OpenSCAD and go straight for build123d which operates on real solid models (BREP): https://build123d.readthedocs.io/en/latest/index.html

I recently designed an eval to see if LLMs can produce usable CAD models: https://kerrickstaley.com/2026/02/22/can-frontier-llms-solve...

Claude 4.6 Opus and Gemini 3.1 Pro can to some degree, although the 3D models they produce are often deficient in some way that my eval didn't capture.

My eval used OpenSCAD simply due to familiarity and not having time to experiment with build123d/CadQuery. There is an academic paper where they were successful at fine-tuning a small VLM to do CadQuery: https://arxiv.org/pdf/2505.14646