This paper proposes an innovative automated multiphysics Microelectromechanical systems (MEMS) co-optimization platform integrating the embedded reduced-order modeling offered by Coventor MEMS+ with MATLAB’s numerical computing environment, addressing critical limitations in conventional design workflows. Our framework demonstrates three key advances. First, the platform enables systematic co-optimization of structural parameters (comb geometries and suspension beams) through constrained design space exploration, achieving an order-of-magnitude improvement in sensor linearity while maintaining baseline sensitivity as validated by accelerometer case studies. Second, the implementation of manufacturing-aware optimization incorporates process tolerance constraints and geometric feasibility checks, effectively bridging the gap between simulation-based optimization and physical fabrication requirements. Third, comprehensive algorithm benchmarking reveals that the modified Nelder–Mead method achieves a superior convergence efficiency (~100× computational efficiency) compared with evolutionary algorithms, while maintaining design quality, providing critical advantages for rapid MEMS prototyping. The platform establishes a new paradigm for MEMS co-design through tight integration of multiphysics simulation, manufacturing constraints, and intelligent optimization algorithms.