For the problems of the unbalanced capability between global search and local search of standard cuckoo search (CS) algorithm, a gradient-based adaptive quick cuckoo search (GBAQCS) is proposed. The direction of the step is determined based on the sign of the gradient of the function. On the one hand, the adaptive search strategy is used to balance the global search and local search capability. On the other hand, the current-guided search method is adopted to improve the convergence precision and rate. The simulation experiments show that GBAQCS fully utilizes and balances the global search and local search capability, and greatly improves the convergence speed and quality of solutions compared with other optimization algorithms.