Medium-range forecasting experiments were conducted using the 0.125-degree weather forecast model configuration of the domestically developed Global-to-Regional Integrated forecast SysTem model (GRIST). The precipitation forecast performance for the baseline version of GRIST was evaluated by comparing with the ERA5 reanalysis data, satellite observation data (GPM) and two global numerical weather prediction models. In addition, sensitivity of GRIST to different dynamic configurations was explored. The results show that GRIST under cold start can simulate the global 500hPa circulation well. Its performance on 500hPa geopotential height anomaly correlation coefficient (ACC500) is comparable to that of the Global Forecast System(GFS) of the National Centers for Environmental Prediction(NCEP). In terms of precipitation simulation, GRIST reproduced the spatial distribution of global mean precipitation overall consistent with the observation. With the increase of integration time, the model presents larger systematic precipitation wet biases than NCEP-GFS over the intertropical convergence zone (ITCZ) and the south slope of the Tibetan Plateau. Based on the analysis of precipitation intensity and frequency, these wet biases are likely due to the overestimation of precipitation frequency. Six key regions are selected to investigate the forecasted precipitation intensity-frequency spectrum and its diurnal variation. The intensity and frequency structure of “heavy precipitation” is better simulated by GRIST than that in NCEP-GFS. The simulation performance of diurnal variation of precipitation is generally reasonable, but an overestimation and advance of precipitation peak was found in several areas. The hydrostatic and non-hydrostatic dynamical cores of GRIST are highly consistent in the 0.125-degree resolution weather prediction, and the vertical 60 layers experiments have certain added value on simulating circulation and precipitation compared with the 30 layers.