The Fengyun-3E (FY-3E), which is the first early-morning-orbit meteorological satellite, was launched by China on July 10, 2021. One of its 11 payloads is the GNSS Occultation Sounder II (GNOS-II), which is a combination of Global Navigation Satellite System (GNSS) Radio Occultation and GNSS reflectometry (GNSS-R) sensor. An important scientific goal of GNOS reflectometry (GNOS-R) is soil moisture detection. Different from previous GNSS-R missions, such as CYGNSS (Cyclone Global Navigation Satellite System), the coverage of FY-3E GNOS-R is pan-global (±85◦ latitude), providing an opportunity for global soil moisture estimation. Owing to data volume limitations, soil moisture retrieval using artificial intelligence is not employed; instead, soil moisture estimation is based on physical scattering models. The LAnd surface GNSS Reflection Simulator (LAGRS) is a spaceborne GNSS-R simulator designed specifically for FY-3E GNOS-R, which provides corresponding theoretical values of surface reflectivity. To ensure the accuracy of soil moisture estimation, the retrieval process includes not only calibration but also removal of surface roughness and vegetation effects. Using the LAGRS model, we obtained the calibration factor and roughness–vegetation factor, and with the mask of Fresnel reflectivity, we achieved soil moisture retrieval. During the entire retrieval process, The Soil Moisture Active Passive (SMAP) ancillary global soil moisture data were treated as reference values. A reasonable degree of consistency was found between the FY-3E GNOS-R soil moisture retrievals and the SMAP data (correlation coefficient: 0.9599, root mean square error: 0.0483 cm3/cm3). This study represents the first illustration of global soil moisture retrieval based on physical models using FY-3E GNOS-R.