Maximum efficiency per ampere (MEPA) control, also known as loss minimization (LM) control, aims to minimize losses in permanent magnet synchronous machine (PMSM) drive systems by optimizing the dq-axis current references. Addressing the complexities associated with precise drive loss models and the challenges of implementing online optimization for MEPA/LM, this article introduces an enhanced model predictive torque control (MPTC). This new strategy incorporates a reformulated cost function that integrates LM control within the conventional MPTC framework, treating it as a constrained optimization problem. The stability of the proposed MPTC approach is theoretically analyzed using Lyapunov stability theory. Simulation and experimental results validate the control accuracy, dynamic response, and efficiency gains facilitated by this innovative strategy.