The AlGaN/GaN HEMT models can be classified into three main groups: the artificial neural network (ANN)-based model, the physical model and the empirical model. The empirical model employs empirical equations to describe nonlinear device elements like the gate capacitance and the drain current. On the other hand, the physical model specifies the characteristics of a device using physical equations or theory. The neural network (ANN)-based model is one more kind of GaN HEMT model. It employs neural networks (NNs) to precisely characterize nonlinear elements like HEMT gate capacitances and nonlinear drain current. Accuracy and outstanding nonlinear fitting capability are the major benefits of ANN-based models. Recently, AlGaN/GaN HEMTs have demonstrated the highest Johnson figure of merit (JFOM), increased frequency response (higher fmax), high power density, high breakdown voltage, best power device figure of merit and low dielectric constant, which makes it into a highly promising transistor technology for multiple applications, including the existing 5G/communication systems, satellite communications, next generation of 5G networks, military and commercial and avionics.