Aiming at the optimal dispatch problem of distribution system with demand response (DR), this paper analyzes the challenges in DR modeling from the perspective of the cyber-physical-social system, i.e., finite data information at cyber level, bounded rationality at social level and source-load interaction characteristic in the physical level. Thus a hybrid model and data driven DR modeling method is developed, which taking the limited data information and the bounded rationality into consideration. Furthermore, a Stackelberg game-based optimal dispatch model is proposed to represent the interaction behavior between distribution network operators (DNO) and electricity consumers, and the proof of the existence and uniqueness of Stackelberg equilibrium is given. Finally, the actual data and electricity price of one province power grid in China are adopted for simulation study. The simulation results show that the proposed DR modeling method can well reflect the user’s actual response behavior and the Stackelberg game-based optimal dispatch approach is helpful to the implementation of real-time price mechanism compared with the time of use mechanism, which plays an important role in reducing DNO and user costs and promoting the utilization of renewable energy.