The shuffled flog leaping algorithm for optimization in function easily falls into local optimal solution and the premature quickly converges of such shortcomings. Combined with the excellent characteristics of cloud model transformation between qualitative and quantitative, an adaptive grouping chaotic cloud model shuffled frog leaping algorithm is proposed based on the cloud model theory. The population is initialized through reverse learning mechanism, the cloud model algorithm is used to local refinement in the region of convergence in order to explore the better position, and the chaos theory is used to obtain global optimization in the space outside the convergence region in order to explore the global optimum position. The simulation results show that the proposed algorithm has fine capability of finding global optimum, especially multi-peak function.