Accurate simulation of wind speed is of significant guiding importance for the future development layout of new energy sectors such as wind and solar power. However, current global climate models exhibit limitations in their ability to simulate wind speeds. Based on CN05.1 high grid observations from national benchmark stations, this study evaluates the capability of ten Global Climate Models (GCMs) from the Coupled Model Intercomparison Project Phase 6 (CMIP6) in simulating the spatiotemporal distribution characteristics of 10-meter wind speeds over mainland China during 1961-2014. Results show that all ten CMIP6 GCMs can capture the climatological seasonal variation of 10-meter wind speeds over China, with maximum speeds in spring and minimum speeds in autumn during the period from 1961 to 2014. However, most models exhibit a systematic overestimation throughout the year, particularly the BCC-CSM2-MR and FGOALS-f3-L models. These ten CMIP6 models are also capable of reproducing the spatial distribution characteristics of the interannual variability of annual and seasonal mean wind speeds over China, although the simulated values are generally smaller than the observed values. In simulating the decadal trends of 10-meter wind speeds from 1961 to 2014, only some of the ten CMIP6 GCMs could capture the weakening trend of annual and seasonal mean 10-meter wind speeds over most regions of China during this period. Moreover, the simulated weakening trends are significantly smaller than the observed trends, and over some regions, the models even fail to simulate the weakening trend, instead showing an increasing trend, which is contrary to the observations. The uncertainties in historical climate simulations also lead to considerable uncertainties in future climate change projections. Therefore, to obtain more reliable future scenario projections, it is essential to apply bias corrections to the model simulations and projections.