<p>Utilizing a slag viscosity model to obtain viscosity directly eliminates the need for costly and time-consuming experimental measurements, significantly improving production efficiency. However, existing models inadequately describe and account for Al³⁺ structures, making it difficult to address the challenge of viscosity prediction when slag composition shifts toward high Al₂O₃ levels, particularly in predicting non-linear behavior. Based on the Reddy–Lv model framework, this study incorporates molecular dynamics (MD) simulations to construct a precise function relating slag composition to structural characteristics. The present model quantitatively describes the relationships between slag composition and the three types of bridging oxygen bonds (Si–O–Si, Al–O–Al, and Si–O–Al) in the CaO–SiO₂–MgO–Al₂O₃ quaternary system. While maintaining high predictive accuracy, the model successfully reproduces the non-linear variation of viscosity with changes in the Al₂O₃/SiO₂ ratio. This work fills an important gap in the current modeling approaches to slag viscosity.</p>