Supervisory Model Predictive Control is one of the technologies that enables an optimal and consistent operation of glass processes, ranging from melter, refiner, and forehearth to product forming. Integrating super-fast non-linear prediction tools into the control loop have resulted in unprecedented level of manufacturing optimisation. This article describes a typical application of a supervisory Model Predictive Control on a float glass melter and the possibilities of fully controlled product changeover on a container glass forehearth. Applications that result in more stable crown and bottom temperature in melters and forehearths, faster recovery after process upsets, reduced product changeover time and improved operator performance.