Preprint reports LLM-guided printability system results from physical trials

A preprint describes an LLM-based system that combines STL geometry evidence with structured material and printer knowledge to recommend printability, materials, process settings and design guidance before fabrication. The researchers report 75.0% printability across 96 physical validation trials and 88.9% task suitability among successfully printed samples. In focused benchmark scenarios, the framework raised Gemini 2.5 Flash-Lite material-selection accuracy from 37.5% under a pure-LLM approach to 90.0%.