Polysulfone (PSF) based mixed matrix membranes (MMMs) are one of the most broadly studied polymeric materials used for CO2/CH4 separation. The performance of existing PSF membranes encounters a bottleneck for widespread expansion in industrial applications due to the trade-off amongst permeability and selectivity. Membrane performance has been postulated to be enhanced via functionalization of filler at different weight percentages. Nonetheless, the preparation of functionalized MMMs without defects and its empirical study that exhibits improved CO2/CH4 separation performance is challenging at an experimental scale that needs prior knowledge of the compatibility between the filler and polymer. Molecular simulation approaches can be used to explore the effect of functionalization on MMM's gas transport properties at an atomic level without the challenges in the experimental study, however, they have received less scrutiny to date. In addition, most of the research has focused on pure gas studies while mixed gas transport properties that reflect real separation in functionalized silica/PSF MMMs are scarcely available. In this work, a molecular simulation computational framework has been developed to investigate the structural, physical properties and gas transport behavior of amine-functionalized silica/PSF-based MMMs. The effect of varying weight percentages (i.e., 15–30 wt.%) of amine-functionalized silica and gas concentrations (i.e., 30% CH4/CO2, 50% CH4/CO2, and 70% CH4/CO2) on physical and gas transport characteristics in amine-functionalized silica/PSF MMMs at 308.15 K and 1 atm has been investigated. Functionalization of silica nanoparticles was found to increase the diffusion and solubility coefficients, leading to an increase in the percentage enhancement of permeability and selectivity for amine-functionalized silica/PSF MMM by 566% and 56%, respectively, compared to silica/PSF-based MMMs at optimal weight percentage of 20 wt.%. The model's permeability differed by 7.1% under mixed gas conditions. The findings of this study could help to improve real CO2/CH4 separation in the future design and concept of functionalized MMMs using molecular simulation and empirical modeling strategies.