This study compares five global machine learning-based weather prediction (MLWP) models (Pangu-Weather, FourCastNet v2, GraphCast, FuXi, and FengWu) using identical ERA5 initial conditions in Eastern Asia and the Western Pacific from June to November 2023. FengWu showed the best performance, followed by FuXi and GraphCast, with FCN2 and Pangu-Weather ranking lower. A multi-model ensemble improved performance, rivaling FengWu. FengWu had the most accurate typhoon track predictions but the largest intensity errors.
