An adaptive grouping difference firefly algorithm is proposed to solve the continuous space optimization problem. In the algorithm, the population is divided into subgroups for optimization based on adaptive grouping strategy and the parametors are adjusted based on the uniform design theory the stochastic perturbation behavior of optimal individuals is updated by using the cloud model algorithm. The updating method of the non-optimal individual is improved by introducing the individual energy efficiency attraction. Finally, the individual variation is completed by using the differential mutation algorithm and chaos theory. Simulation results of the typical complex functions show that the algorithm has good the convergence precision and computing speed.