In a steel melting shop, liquid steel temperature is the foremost parameter for casting quality products and uninterrupted operation. In this paper, reliable machine learning (RML) framework has been proposed for predicting temperature at a basic oxygen furnace (BOF), ladle heating furnace (LHF/LF), Ruhrstahl Heraeus (RH) degasser, and caster (CCM). RML is based on the stacking of eight machine learning models. Feature selection is done by application of the variance inflation factor. Prediction accuracy of the proposed model at BOF turndown, LHF opening, LHF despatch, RH despatch, and CCM is 95%, 99%, 96%, 99%, and 99%, respectively. Prediction of temperature within the tolerance limit will reduce the time and cost of temperature measurement. Additionally, an analysis of the correlation between parameters and temperature is done. This will enable operators to optimize steelmaking process to meet the target temperature. The application of the present work will accelerate the efficiency of industry 4.0.