High-lift airfoil design is subjected to many constraints in terms of regulation and efficiency. It also involves complex flows that are still challenging to be estimated by Computational Fluid Dynamics. The time required to solve flow around a multielement airfoil makes it difficultly applicable to optimization. However, from legacy, approximated or empirical methods are available to engineers to quickly estimate performances of such designs. But those methods are limited in their range of applicability and accuracy. This paper presents a multifidelity optimizer using two models for the fluid dynamics: a low-fidelity that is used under a trust region scheme and a high-fidelity used to validate candidate points. The lowfidelity function is corrected by a Radial Basis Function surrogate model interpolating the error between the highand low-fidelity models at the sampling points. Points are sampled during the optimization and therefore the method does not require (but can use) any sampling strategy prior of the optimization. The method uses an implementation of the Tabu Search that provides non-deterministic behavior for global exploration. The method is applied to a simple low speed single airfoil case and then to a multielement airfoil case provided by the industry. The results show reduction in terms of number of high-fidelity function calls and in terms of time for the convergence for both test cases. The variability of the results, more important for the multielement airfoil, needs to be addressed and is currently under work.