Purpose <p>Physical liver models are essential for surgical training and biomechanical research, offering standardized, reproducible, and ethically compliant experimental platforms. However, accurately replicating the nonlinear mechanical behavior of real liver tissue remains a challenge for existing simulant materials.</p> Methods <p>This study proposed a design strategy for particle-reinforced hyperelastic composites based on an improved Mori–Tanaka method. An incremental constitutive model under finite deformation was established to theoretically predict the nonlinear stress–strain response of the composite material. Using Ecoflex silicone as the matrix and Dragon Skin microspheres as the reinforcing phase, composites with volume fractions ranging from 10 to 30% were fabricated and subjected to quasi-static uniaxial compression tests to validate the model. By leveraging stress–strain data from both bovine and human livers, inverse optimization was employed to determine the optimal material formulations that match the mechanical behavior of liver tissues.</p> Results <p>The mean relative errors between the model predictions and experimental results were 16.11, 16.98, and 10.86% for volume fractions of 10, 20, and 30%, respectively, validating the accuracy of the theoretical model. The composites with an EF10-30% matrix and a reinforcement volume fraction of 16.14%, as well as those with an EF10-50% matrix and volume fraction of 19.37%, closely matched the mechanical properties of bovine and human liver tissue, respectively.</p> Conclusion <p>This study establishes the quantitative framework for designing particle-reinforced liver simulant material, integrating micromechanical modeling. The methodology achieves precise matching of nonlinear tissue mechanics and provides systematic guidelines for developing high-fidelity simulants for surgical training and biomechanical applications.</p>