<p>Magnetorheological polishing (MRP) is a precision finishing technique that enhances surface quality by utilizing the controllable properties of magnetorheological (MR) fluids. Under a magnetic field, the fluid forms a flexible polishing layer that conforms to complex surfaces. This study proposes an optimized magnetic field generation component (MFGC) that accommodates variable working gaps, aiming to increase magnetic flux density and yield stress, thereby improving polishing performance. A computational fluid dynamics (CFD) numerical model was developed to evaluate magnetic field distribution and yield stress under various working gap conditions. The results indicate that the average magnetic flux density increased from 0.44 to 0.67&#xa0;T, while the average yield stress rose from 11.7 to 21.2 kN/m². A multi-objective genetic algorithm (MOGA) was employed to optimize the MFGC, leading to a 1.48-fold increase in magnetic flux density and a 1.81-fold increase in yield stress compared to the non-optimized design. Experimental validation on SKD11 spherical workpieces confirmed a reduction in surface roughness (<i>Ra</i>) from 120 to 18&#xa0;nm at a 1.6-mm working gap and a decrease in shape deviation from 4.73 to 1.27&#xa0;µm at a 2.0-mm working gap. These findings demonstrate that, with an optimized MFGC, the MRP process offers high-efficiency finishing for complex geometries.</p>