The Imaging/Sounding Microwave Radiometer–Improved (MTVZA‐GY) on board the Russian meteorological satellite, Meteor‐M N2‐2, launched in 2019, provides daily observations of Earth's atmosphere and surface from a polar orbit. Here, its performance in a numerical prediction model – the Global/Regional Assimilation and Prediction System–Global Forecast System (CMA_GFS), which involves the Advanced Radiative Transfer Modeling System (ARMS) – was evaluated. After supplementing some lacking information during data preprocessing, the characteristics of all available channels (24 in total) were evaluated by comparison among channels, with background fields, and with similar active instruments in CMA‐GFS, as well as between different radiative transfer models. Failed calibration was found in all window channels. Scan position biases, ascending/descending biases, and striping noises were widely discovered in temperature‐sounding channels, as well as larger biases in humidity‐sounding channels. Following quality control and bias correction, only two temperature‐sounding channels were feasible for assimilation into CMA‐GFS within the observational errors calculated by the a posteriori verification scheme. A one‐month experiment confirmed that these two channels have positive impacts on the analysis of both thermal and dynamic fields, as well as short‐term weather forecasting in the Northern Hemisphere and tropics. Short‐term global forecasting of moderate rainfall was also improved. This work is a pioneering attempt at examining the potential and impacts of assimilating MTVZA‐GY in a numerical weather prediction model system. It also provides guidance for the manufacture and usage of the instruments that will be on board the three satellites planned for launch by the Russian Federation in the next three years.