The industrial importance of air bending applications with high-strength steels has increased in recent years. However, the limited cold formability of high-strength steels requires the application of punches with large radii. The usage of large radii tooling leads to a change of the loading scheme from traditional 3-point to 4-point bending. Therefore, the conventional formulas used for the prediction of 3-point bending operations fail to deliver accurate results. In the current work, two regression approaches are proposed to achieve an accurate prediction of industrially important bending parameters. The first model is based on phenomenological observations and fitting of well-established formulas. The second model utilizes the circular approximation model as base for the regression prediction. Large radius bending is used mostly for high-strength steels. Therefore, Strenx 700 MC and Strenx 1300 with thicknesses of 4 and 6 mm were used for the investigation. The following bending characteristics have been studied: position of contact points, bend allowance, bending force and springback. The predicted values are compared with the experimental results to assess the prediction accuracy of the proposed methods. Neither the phenomenological, nor the circular model delivers superior overall prediction quality for all output parameters, thus a combination of the two regression models is an attractive option for a better prediction of large radius air bending.To be able to reduce the post-forming operations, an accurate prediction of any forming process is required. A number of methods can be used for that purpose, but in this work, we focus on regression