Independent dose calculations (IDC) for quality assurance in proton beam therapy treatment planning become more and more interesting these days. As opposed to quality assurance (QA) measurements, IDCs free up valuable beam time for patient treatment and research and can help to increase the treatment efficiency of a therapy facility. A popular method for precise independent dose calculations are Monte Carlo (MC) simulations. Before the actual IDC can be performed a beam model is necessary to fully exploit the accuracy of MC simulations. The aim of this thesis was to find a beam model for the research beamline at MedAustron for the MC toolkit GATE. Since beam modeling is usually performed manually and is thus a time-consuming and tedious task, a novel optimization tool was used that automated the process of finding a beam model. Beam modelling was split into two parts: beam energy modeling and optical beam modelling. For beam energy modeling, proton beams (62.4, 97.4, 148.2, 198.0, 252.7 MeV) stopping in a 40x40x42 cm2 water phantom were simulated using GATE v8.1. Resulting simulated Bragg curves were analyzed with regard to range and Bragg peak width and fitted to measured Bragg curves. Corresponding measurements were performed using an MP3-PL water phantom (PTW, Freiburg, Germany) incl. Bragg peak chambers. For optical beam modeling, proton beams (with same energies as for beam energy modeling) traversing phase-space actors at selected isocenter distances (ISDs) were simulated using GATE v8.0. The FWHM (full width at half maximum) of the transverse intensity profiles was extracted and the simulated FWHM-vs-ISD curves were fitted to measured curves. Corresponding measurements were per- formed using a Lynx PT detector (IBA dosimetry, Schwarzenbruck, Germany). Using the obtained beam model as basis, a special QA box treatment plan in water from the RayStation (RaySearch Laboratories, Sweden) treatment planning system (TPS) was recalculated in GATE. The resulting dose distribution was com- pared to the dose distributions predicted by the TPS. The optimized beam parameters showed clinically acceptable agreement when val- idated with the respective measurements. Maximal deviations for Bragg curve ranges were less than 0.1%. Deviations in Bragg peak width reached up to 6%. Most absolute deviations for the FWHM lay within +0.5 mm and -0.5 mm. The corresponding relative deviations for the FWHM lay within +4% and -6%. Comparison of the recalculated dose distribution in GATE with the TPS-predicted distribution showed relative deviations in dose of less than 2.5% for most cases. Therefore, the obtained beam model can be used as the basis of an independent dose calculation tool for the research beam line at MedAustron.The optimization tool allowed to automatically find an accurate beam model within 35 to 40 hours on a conventional notebook. Employment of the tool at MedAustron for carbon ions is planned where it will help to reduce the amount of try-and-error based beam modeling sessions. Current work in progress deals with testing the automated beam modeling approach also at other particle therapy centers.