Electropermanent magnet (EPM) composites integrate permanent magnets and iron at the microscale, enabling magnetic field generation without external current while allowing control of the magnetic field intensity. Owing to these characteristics, EPM composites have attrac-ted considerable attention as promising materials for electromagnetic devices, including magnetic actuators. Since the performance of EPM composites strongly depends on microstruc-tural design, multiscale topology optimization, which links the microscale and macroscale, has been proposed as an effective approach for designing EPM composites [1]. However, existing multiscale topology optimization studies on EPM composites have primarily relied on homogenization models that assume linear magnetic behavior of iron. In practice, electromagnetic devices operate in nonlinear regimes dominated by magnetic saturation of iron [2], and neglecting such nonlinear magnetic behavior makes it difficult to obtain designs that accurately reflect actual operating conditions.In this study, a multiscale topology optimization framework for magnetic actuators considering the nonlinear magnetic behavior of EPM compos ites is proposed. The proposed multiscale topology optimization procedure consists of three main steps. First, a surrogate model of effective material properties accounting for the nonlinear magnetic behavior of EPM composites is constructed. Second, multiscale topology optimization is performed for magnetic act-uator design using the developed surrogate model. Finally, the optimized EPM microstructure is reconstructed at the macroscale.The effectiveness of the proposed method is examined through numerical examples. In the numerical examples, the validity of the nonlinear homogenization model is verified, and the multiscale topology optimization results for magnetic actuators ob tained using the proposed framework are presented.