This project is exploring the development of new methods for building large multimodal generative artificial intelligence models and demonstrating neural methods on a mission-relevant exemplar. The models developed here will play a foundational role for Sandia developed multimodal models. BANYAN will demonstrate the approach by solving problems that were previously infeasible in the nuclear deterrence domain.