In the production of high-quality steels, the removal of non-metallic inclusions is crucial. This paper presents a slag modeling approach to investigate the influence of slag composition on the removal of inclusions in Mn-Si killed steel. The study utilized the Spark-DAT technique to measure inclusion types and amounts in different stages of the steelmaking process. MnS, CxS, and CaS were identified as the main inclusions, with their amounts increasing from the ladle furnace to continuous casting, contrary to the desired reduction. Potential reasons for this increase include inadequate desulfurization, inclusion reversal, and slag carryover. The analysis revealed that the initial slag composition in the ladle furnace better explains inclusion variation compared to the final composition. Correlation analysis indicated that EAF slag data correlated more with CaS and SiCa inclusions, while initial ladle furnace slag data correlated better with MnS inclusions. Regression analysis highlighted the influence of slag viscosity on CaS inclusions and FeO activity on CaSi inclusions. Effective inclusion removal can be achieved by controlling FeO activity and optimizing slag viscosity, emphasizing the significance of slag composition and process parameters in enhancing steel quality.