Considering the local magnetic characteristics of surface-mounted permanent magnet (SPM) motors, the paper proposes an adaptive connecting equivalent magnetic network (ACEMN) model to accurately predict SPM motor performance. First, for modeling the magnetic field at the inclined boundary of the stator pole shoe, a diagonal hybrid permeance element covering two materials is developed. And considering the parallel magnetization of PMs, a branching calculation of the magnetomotive force source is performed inside a cross-shaped permeance of a fan-shaped mesh. Then, by analyzing the phenomenon of magnetic field line deflection at the air gap boundary, an air gap node connecting way based on adaptive conversion of connecting permeances is built. Thereby, the rotating magnetic field of the air gap can be accurately described using the different connecting permeances with variable size. To accelerate the nonlinear solution for saturated element permeability, a hybrid iterative method is used. The validity of this modeling method is verified by finite element analysis (FEA) and prototype experiments, which allows a satisfactory compromise between accuracy and calculation speed.