In this study, the finite-failure NHPP software reliability model was applied to the software development model based on the Weibull lifetime distribution (Goel-Okumoto, Rayleigh, Type-2 Gumbe), which is widely used in the software reliability field, and then the cost attributes were compared and evaluated. For this study, failure time data detected during normal operation of the software system were collected and used, the most-likelihood estimation (MLE) method was applied to the parameter estimation of the proposed model, and the calculation of the nonlinear equation was solved using the binary method. As a result, first, in the software development model, when the cost of testing per unit time and the cost of removing a single defect increased, the cost increased but the release time did not change, and when the cost of repairing failures detected during normal system operation increased, the cost increased and the release time was also delayed. Second, as a result of comprehensive comparative analysis of the proposed models, it was found that the Type-2 Gumble model was the most efficient model because the development cost was lower and the release time point was relatively faster than the Rayleigh model and the Goel-Okumoto basic model. Third, through this study, the development cost properties of the Weibull distribution model were newly evaluated, and the analyzed data is expected to be utilized as design data that enables software developers to explore the attributes of development cost and release time.