Having 6500 integrated sensors and a wide network of embedded systems, the ADREAM building can be categorized as one of the prototype smart buildings of France. As needs for intelligent management of energy are growing, the associated project provides a multidisciplinary platform of experimentation, developing solutions for efficient Energy Networks, HVAC Systems, Photovoltaics, and Smart Grids. This paper provides the overview of the project along with three different modeling techniques, illustrating their strengths and limitations on simulating the thermal behavior of the building and the functioning of the different energy systems. A thermal model of the building was developed and calibrated for energy consumption analysis and prediction, using the software Pleiades+Comfie. A "black box" model was developed using artificial neural networks for the simulation of energy system parameters and the exploration of efficient control strategies. The last method provides the overview of developing an allinclusive electrically equivalent physical model with Matlab/Simulink for simulating the global functioning of the building and its HVAC systems. The conclusion addresses the utility of exploring and combining different types of models for optimizing the energy management of Smart Buildings.