Abstract: Multi-fidelity aerodynamic modeling, which integrates experimental and simulation data of varying fidelity levels, effectively enhances computational efficiency and prediction accuracy in analyzing aircraft aerodynamic characteristics. Traditional multi-fidelity modeling methods typically assume ‌a single type of mapping relationship (either linear or nonlinear)‌ between high- and low-fidelity aerodynamic data. However, practical analyses under complex flight conditions, such as transonic flows and high-angle-of-attack scenarios, reveal that multi-fidelity data often exhibit hybrid correlations ‌that combine‌ both linear and nonlinear features. To better address these coexisting ‌complex linear-nonlinear interdependencies‌ across fidelity levels, the present study proposes a novel ‌Multi-fidelity Gaussian Process Regression (MFGPR)‌ model based on the ‌Nonlinear Autoregressive Gaussian Process (NARGP)‌ framework. By integrating linear and nonlinear kernel functions, the MFGPR extends the capabilities of NARGP to simultaneously capture ‌both‌ intricate nonlinear relationships and linear dependencies inherent in multi-fidelity datasets. To validate the effectiveness of MFGPR, two analytical functions were tested and compared with conventional multi-fidelity modeling approaches. The results demonstrate that MFGPR achieves performance comparable to ‌CoKriging‌ in handling linearly correlated problems, ‌while exhibiting‌ superior prediction accuracy and significant efficiency advantages ‌in addressing‌ complex nonlinear correlations. Finally, the application of MFGPR to two engineering cases further illustrates its potential and outstanding performance in aerodynamic ‌force modeling‌ applications.