Recent developments integrating micromechanics and neural networks offer promising paths for rapid predictions of the response of heterogeneous materials with similar accuracy as direct numerical simulations. In this presentation I will present recent extensions to the deep material network (DMN) to predict the thermo-elasto-plastic response of composite materials and compare these predictions to Fast Fourier transform direct numerical simulations. This approach is based on a microstructure-aware binary tree-type network with connected mechanistic building blocks. These building blocks use analytical homogenization solutions to describe the overall material thermos-mechanical response of a fixed microstructure. In the offline training mode, the network is trained on coefficient of thermal extension and elastic stiffness data. In the online prediction mode, the network does not need to be trained on data and is able to extrapolate to predict the thermo-elasto-plastic response of composites subjected to various thermal boundary conditions. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy National Nuclear Security Administration under contract DE-NA0003525.