<p>This study presents a stepwise thermodynamic equilibrium model to predict slag evolution during the argon–oxygen decarburization (AOD) process in stainless steelmaking. The model incorporates key process parameters, including gas composition, pressure, reductant type, flux ratios, and temperature, across the decarburization, reduction, and cooling stages. Thermodynamic predictions were validated against experimental data, including XRD and EPMA–WDS analyses of industrial slag samples. The model enables optimization of the decarburization stage by allowing stepwise variations in gas composition, temperature, initial steel chemistry, and the average ferrostatic pressure within the furnace. It accurately captures the dynamic phase transformations occurring in the slag, including the formation of MgCr₂O₄ and CaCr₂O₄ during decarburization, as well as bredigite, merwinite, and cuspidine under different CaO/SiO₂ and MgO/SiO₂ ratios during reduction. Consequently, the model provides a consistent explanation of slag phase evolution throughout the entire process. It also reveals the influence of reductant strategy (Si <i>vs</i> Si–Al mixtures) and flux composition on phase stability, Cr retention, and slag homogeneity. Characterization studies show that, during decarburization, CaO reacts with Cr₂O₃ to form a CaCr₂O₄ layer on the surface of the lime particles. Additionally, rapid cooling after tapping can kinetically hinder the full transformation of equilibrium phases, resulting in the persistence of metastable β-C₂S. By accurately predicting slag behavior under industrial conditions, the model supports process optimization, improved Cr recovery during decarburization, reduced flux consumption, and overall, more efficient, sustainable stainless steelmaking operations.</p>