To guarantee the individual benefits of different system operators in the optimal power flow of transmission-distribution-coupled (TDC) systems, a generalized Nash game (GNG) model is proposed, under which transmission system operators (TSOs) and distribution system operators (DSOs) optimize their non-cooperative objectives in a coordinated manner. Then, an improved iterative method with the successive-intersection approximation is proposed to solve the GNE to achieve generalized Nash equilibrium (GNE) points. Further, to reduce the unnecessary costs led by the GNG, a GNE-based Pareto solution is achieved with the heterogeneous decomposition algorithm (HGD). Numerical experiments show that the results under GNE are significantly different from those under the cooperative model. Particularly, the individual benefits of DSOs are protected. Also, the GNE-based Pareto solutions can effectively reduce unnecessary costs and improve the overall benefits compared with the results under GNE. Besides, the proposed improved iterative method for solving the GNE shows better convergence and efficiency compared with the conventional simple fixed-point-iteration-based method.