In order to study the effect of different types of fibers on the bending properties of geopolymer concrete (GPC), a series of fiber‐reinforced geopolymer concrete (FRGPC) samples were prepared, and their flexural properties were evaluated by four point bending test and bending toughness evaluation index. Then, based on the fiber characteristic value (λ), a new bending strength prediction model and a three‐dimensional (3D) curved surface prediction model of the bending toughness of FRGPC are proposed and fitted. Finally, a new method for the selection of fibers in GPC is designed by calculating the derivative of the quadratic relationship between fiber reinforcement factor (S) and λ. The results show that all types of fibers have a certain effect on improving the bending strength of GPC samples, and the lifting rate is end‐hook steel fiber (SF), basalt fiber (BF), and short polyethylene polypropylene fiber (SPPF), from high to low. Mixing the above types of fibers with long polypropylene fibers (LPPF) will produce better performance. Among them, the hybrid fiber‐containing SF has the best performance. The bending performance prediction model deduced in this paper has a good correlation coefficient, which can be used to predict the bending performance of FRGPC. By making dS/dλ = 0 can calculate the optimal λ, and this λ can maximize the parameter S and guide the selection of optical fibers.