A model architecture defines how a machine learning model is structured—how data flows through it, how different components interact, and how it makes predictions.