The aim of the work is to identify the influence of the chemical composition of steels and special-purpose alloys on the formation of their physicochemical and structural-sensitive properties. This problem is solved by mathematical modeling of the inseparable chain «composition - structure – property» taking into account the parameters of interatomic interaction in the melt based on the concept of a directed chemical bond. A steel melt is considered as a chemically homogeneous system, and the state of the melts is expressed through a set of integral parameters, the main of which are: Zy - system charge state parameter (e); r - statistically average internuclear distance (10-1nm); tgα is a constant for each element, which characterizes the change in the radius of the ion as its charge changes. On the basis of experimental information on properties and using the parameters of interatomic interaction, computational models are proposed for predicting the properties of steels and alloys. The forecast models took into account the parameters of micro-inhomogeneity of steel, which ensured a high accuracy of the operational forecast. A comparative analysis of the results of steel melting with the corresponding calculations based on the JMatPro software package confirmed the effectiveness of using the interatomic interaction parameters as models. The proposed models for determining the melting of chromium-nickel steels are recommended for use with the content of basic elements Cr, Ni from 0 to 30%. The research results are recommended for use in industrial environments through the integration of the developed models in the process control system of steelmaking, which will contribute to the directed formation of the composition and properties of smelting products, as well as reducing energy costs.