Abstract: In order to improve the forecast capability of quantitative precipitation products in Panxi Region, the precipitation forecasts in summer 2021 of Grid Weather Forecasting (GWF), South West Center-WRF ADAS Real-time Modeling System (SWCWARMS), European Centre for Medium-Range Weather Forecasting (ECMWF) and China Meteorological Administration Meso-Scale Model (CMA-MESO) are tested and analyzed. The results show that: (1) ECMWF has the most obvious bias about the rainy days, but the forecast of rainstorm magnitude is drier than the observation, while the forecast deviations of various magnitudes of other products are mainly wet. The forecast effect of rainy days is better in the south of Panxi region than that in other places which have a bigger false alarm rate. The central part of Liangshan Prefecture has a larger deviation of heavy rain days, while the southern part of Panxi region performs better. (2) According to the 16 process tests, the performance of all forecast products in August is better than that in other months. GWF products above 25 mm has the largest number of processes with TS (Threat Score) score over 20 points, which has the most stable forecast effect, while CMA-MESO model has the largest false alarm rate. (3) The start time of 3h cumulative heavy precipitation for each product is mostly earlier than the observation. The rain intensity of SWCWARMS is too high but it has a better forecast effect of the process with longer duration, while the cumulative precipitation of GWF and ECMWF is smaller than the observation.