The present study addresses the challenge of stabilizing the convergence of topology optimization for elastoplastic materials. Conventional methods often face difficulties in convergence performance due to the discontinuity of the gradient at the yield point in the stress-strain relationship of traditional elastoplastic models. Conventional topology optimization methods, despite relying on gradient-based optimization techniques that require C1 continuity, have historically tolerated such discontinuities, leading to suboptimal results. To overcome this limitation, the proposed approach incorporates the subloading surface model, an elastoplastic material model introduced by Hashiguchi [1]. This model enables a smooth transition from the elastic to plastic state, thereby resolving the discontinuity issue and enhancing convergence stability. By integrating this model into topology optimization, the proposed method aims to achieve reliable and efficient optimization for elastoplastic materials. The objective of optimization is to maximize the energy absorption capacity of a structure under prescribed loading conditions. The sensitivity analysis for the optimization process is derived analytically, building upon the framework established by Kato et al. [2]. Numerical examples are presented to validate the proposed method. These examples demonstrate that the proposed approach improves the convergence performance compared to conventional methods. The findings suggest that the proposed approach with subloading surface model has significant potential to advance the field of topology optimization for not only elastoplastic materials but also other nonlinear behavior.