The traditional model used for computing the costs of post-earthquake reconstruction projects is based on the back-propagation neural network. It requires a complex computation process and provides low convergence efficiency and low-accuracy results. In this work, a model based on the improved genetic algorithm for computing the costs of post-earthquake reconstruction projects is proposed. The cost model is optimized in accordance with the factors that influence the cost of post-earthquake reconstruction projects, and the cost function model is established with superior cost simulation data. The data of the cost function model are analyzed on the basis of the T coefficient, and the data parameters of the cost function model are determined on the basis of the binary calculation law. Then, data parameters with high precision are obtained through formula calculus, and accurate cost data are obtained. Experimental results show that the designed model can accurately and quickly estimate the cost of post-earthquake reconstruction projects.