With the ongoing development of robotics, particularly walking robots and advanced manipulators, alongside the accelerating electrification of manufacturing plants, the diversity of applications and environments in which servo-drives must operate is expanding significantly. Consequently, this increases the requirements regarding accurate trajectory tracking and robustness to parameter variations, while maintaining high efficiency characterized by low current and speed ripples. To address these challenges, this paper proposes predictive control solutions for PMSM servo-drives that ensure high-precision position control and robustness to variations in the moment of inertia, and compares their performance with that of the Cascaded Control Structure (CCS). The authors implement both a novel full and a hybrid scheme of the Fast Gradient Method Model Predictive Position Control (FGM-MPPC) and introduce a novel predictive current control scheme, termed Preliminary Voltage Vector Selecting Model Predictive Current Control (PVVS-MPCC), which generates candidate voltage vectors via a nonlinear function. The investigations were conducted using a variable-frequency sinusoidal reference trajectory under three distinct system inertia values. The research results demonstrate that the application of a hybrid CCS utilizing PVVS-MPCC effectively reduces current, speed ripples while enhancing system robustness. Conversely, the implementation of the full FGM-MPPC provides greater robustness and improved trajectory-tracking accuracy at low speeds.