<p>Current control performance is crucial to SPMSM drive. However, conventional current control methods rely on the physical model of SPMSM and are susceptible to motor parameter mismatch. In order to avoid the problem of parameter mismatch, this paper proposes auto-tuning model-free predictive current control to achieve fast and accurate current control with strong robustness. Firstly, model-free predictive current control with improved extended state observer is proposed in this paper to enhance the rapidity and robustness of current control. Next, stability and tracking error are analyzed in detail which provides theoretical support for the proposed method. Then, the auto-tuning method is proposed to eliminate the dependence on preset nominal inductance, which serves as a start-up self-tuning process before the motor runs. Finally, experiments are conducted on the motor test platform to verify the efficacy of the proposed method.</p>