Software has frolicked an increasing significant role in methods acquisition, engineering, development, mainly for large complex systems. For such systems, accurate estimates of the software costs are a perilous part of effective software management. The preparation of forecasting the cost of the software has evolved, but it is far from perfect. Constructive cost model is the most widely used among all the models available. The proposed model focusses on data set obtained from 91 models nick named as “Maxwell”. The projected model is tested and the test results from the neural network are matched with that of COCOMO model. From the experimental results, it is concluded that the amalgamation of the proposed model for the “Maxwell” data has proved better than the existing COCOMO and another model as obtained from literature.