Conventional model predictive current control is recognized as a control strategy characterized by excellent performance. However, its control performance may produce current harmonics when there is a mismatch in motor parameters. To address this issue, the paper first proposes a model-free predictive current control based on nonlinear extended state observer (NESO) that reduces steady-state current ripple without relying on motor parameters, aiming to effectively mitigate steady-state current ripple even in the presence of parameter mismatches. Moreover, the nonlinear function-based NESO designed in this paper demonstrates minimal observation error. This method utilizes only the system’s input and output, without considering any motor parameters, thereby enhancing the robustness of the control system. Simulation experiments confirm that the proposed model-free predictive current control method for permanent magnet synchronous motors exhibits robustness and excellent dynamic performance. It effectively enhances motor control performance, thus validating the effectiveness of the proposed model-free predictive current control (MFPCC).