GPT-6 Astra redesigns magnesium stent with 1,650 CAD parameters

arxiv.org1 min read

Researchers investigated whether a general-purpose large language model can redesign real engineering components by combining its knowledge with nonlinear finite-element simulations. Using a budget of 15 simulation solves, GPT-6 Astra edited a magnesium stent through a CAD geometry parametrization of roughly 1,650 values and successfully lowered a high percentile of bending strains by 32%. Open article →