Abstract: Hyperspectral satellites can accurately capture the spatial and spectral information about surface features. However, their application performance largely depends on the accuracy of surface reflectance inversion. Based on data from the GF-5 01A and GF-5B satellites, this study conducted surface reflectance inversion and verified its accuracy. Specifically, radiance products were obtained through the preprocessing of level-1 products. Then, surface reflectance inversion of the post-preprocessing radiance products was carried out using a radiative transfer model. The inversion results were subjected to cross-validation between the two satellites based on quantitative metrics, along with comparison validation against the MODIS surface reflectance products. The results indicate that both GF5 01A and GF5B satellites can effectively capture the spectral information of surface features. The surface reflectance inversion based on the hyperspectral data from the two satellites yielded the same level of accuracy, and the inversion results are highly consistent with the MODIS surface reflectance products. Additionally, tests on different types of surface features confirm that GF5 01A and GF5B satellites maintain consistent surface reflectance inversion accuracy across various surface features.