Typhoon Doksuri (2305) maintains as super typhoon at sea for about 72 h, lands in Fujian as severe typhoon, and maintains for 22 h after landing, causing serious wind and rain impacts to China’s offshore, coastal, and inland areas. There are great differences in the forecast of various numerical models, which cause great difficulties to operational forecast. In order to better explain and apply the numerical models, this paper focuses on the verification for the performance of CMA-TYM (China Meteorological Administration_Regional Mesoscale Typhoon Numerical Prediction System) and CMA-GFS (China Meteorological Administration_Global Forecast System) track and intensity prediction during the process of Doksuri, and compares them with the forecasts of ECMWF (European Centre for Medium-Range Weather Forecasts) and NCEP (National Centers for Environmental Prediction，NCEP) to find out the advantages and disadvantages of CMA models. The results show that the CMA-TYM track forecast errors are similar to those of NCEP and larger than those of ECMWF and CMA-GFS. The prediction of its movement direction at the initial stage has a large deviation, which is obviously biased to the right side of the real track. The track forecast starting on 23 June has a good grasp of the landing point, and the error is mainly due to the prediction of faster moving speed. Compared with other models, the CMA-TYM prediction of rapid intensification process has significant advantages. The CMA-GFS track forecast performs well, the error is slightly larger than ECMWF, but smaller than NCEP and CMA-TYM. CMA-GFS can provide a good reference for long-term prediction, while its result is obviously weak in intensity prediction.