Cells are known to sense and respond to the stiffness of their insoluble environment (matrix) [1]. Yet the stochastic nature of cell–matrix interactions complicate experimental quantification of cell-perceived stiffness in 3D matrices. To address this, we developed an efficient in silico approach to quantify cell-perceived stiffness in microporous scaffolds [2]. This approach demonstrated that cell-perceived stiffness depends on scaffold structural parameters, attachment location, contractility, and neighbouring cell-matrix interactions. Our initial model was validated at the macroscopic scale using compressive simulations and assumed that each cell adhered to the scaffold at two points. In comparison, recent models treated cell response to matrix mechanics as a function of bulk matrix stiffness and were validated based on time-lapsed cell behaviour [3,4]. Such models, unlike our approach, described the mechanical environment perceived by individual cells implicitly, and possibly overlooked cell-scale variations in force transmission and attachment. Here, we extend our framework to a coupled model that can quantify matrix deformations and cell-perceived stiffness via the finite element method and simultaneously predict cell response via agent-based modelling. Model predictions of cell response can be validated against experimental observations, establishing the relevance of cell-perceived stiffness in cell mechanosensing and providing a quantitative link between scaffold mechanics and cell behaviour. Additionally, this model generalises our analytical definition of cell-perceived stiffness, to cells that attach to more than two attachment points. In this formulation, each cell is represented as a system of springs whose equivalent stiffness equals the cell-perceived stiffness. We demonstrate that contrasting homogeneous and heterogeneous pore architectures influences cell-perceived stiffness and response, despite similar bulk stiffness, suggesting bulk stiffness alone cannot describe the cellular mechanical environment. Together, these results position the model as a tool for optimising biomaterials and understanding the role of biomechanics in key phenomena observed in physiology, pathology and biomaterial-induced regeneration.