In the model predictive torque control (MPTC) of permanent magnet synchronous motor (PMSM), due to the different dimensions of control variables in the objective function, in order to select the optimal weight coefficient in the cost function, an improved Gaussian mutation pigeon-inspired optimization algorithm (GPIO) is proposed to realize the self-tuning of the weight coefficient. First, the objective function of PIO is designed based on the principle of cost function minimization, and the weight coefficient is evaluated by the root mean square of current id and iq pulse error. Second, based on the traditional Gaussian mutation operator, adaptive dynamic parameters are introduced to realize mutation. The simulation and experimental results show that the improved algorithm is applied to the model predictive torque system of permanent magnet synchronous motor, and the optimal weight coefficient can be set within less iterations, it reduces the torque error and current harmonic distortion rate.