The interest in model predictive direct torque control (MP-DTC) for high-performance dynamic control of electric drives has been growing. Unlike the conventional direct torque control (DTC), MP-DTC can achieve optimal voltage selection by predicting the control variables evaluations using a machine model. However, the model-machine mismatch can degrade the control performance significantly, especially for the interior permanent magnet synchronous machine (IPMSM) because of its highly nonlinear characteristics. Therefore, this paper describes an investigation of the MP-DTC algorithm that exploits the advantages of a finite element analysis (FEA) based model in representing the behavior of the IPMSM precisely, including the magnetic saturation and spatial harmonics effects, to suppress the torque pulsations and eliminate the steady-state torque error. Moreover, this approach optimizes the duty ratio simultaneously with the voltage vector selection to guarantee further torque ripple reduction under steady-state operation, especially at low speeds. Simulation results of an 80-kW IPMSM drive are presented to validate the model and the proposed control method.