Abstract: GRAPES_YY is a global non-hydrostatic model developed on the basis of a spherical quasi-uniform Yin-Yang grid, which is currently capable of forecasting medium-range weather with a full physical parameterization package being implemented. To verify the model performance and its capability of predicting multi-scale weather systems, especially the circulations, precipitation and corresponding error patterns, we first analyze a series of medium-range hindcast from 2 June to 17 July 2018, and then highlight the analysis of a Meiyu process in East China during 4—7 July 2018. The ERA-interim data and gauge-radar-satellite combined precipitation data in China are taken as the "true" situation. The model shows excellent numerical stability after the improvement of the upper boundary condition. The results display realistic distributions of global circulation and precipitation within the first three-day prediction. The model successfully reproduced the two 24 h accumulated precipitation peaks in the low-latitude and mid-latitude regions, respectively, although it overestimated convective precipitation in the low-latitudes and large-scale precipitation in the mid-latitudes. The model not only well described the location, orientation and north-south displacement of the Meiyu rain belt, but also realistically simulated the movement and evolution of the synoptic systems for the case occurred during 4—7 July. However, the simulated heavy rainfall center is weaker than observations and is located further north due to the limitation of resolution and inaccurate description of the wet physical process. In conclusion, the GRAPES_YY model on the Yin-Yang grid produces reasonable results. It demonstrates ability to predict precipitation associated with Meiyu fronts in the mid-latitude and the performance is slightly better than the original GRAPES_GFS model.