This study evaluates the ability of three operational models, AROME, ARPEGE and ECMWF, to predict the boundary-layer turbulent processes and mesoscale variability observed during the Boundary Layer Late-Afternoon and Sunset Turbulence (BLLAST) field campaign. AROME is a 2.5 km limited area non-hydrostatic model operated over France, ARPEGE a global model with a 10 km grid-size over France and ECMWF a global model with a 16 km grid-size. We analyze the representation of the vertical profiles of temperature and humidity and the time evolution of near surface atmospheric variables as well as the radiative and turbulent fluxes for a total of 12 24h-long Intensive Observing Periods. Special attention is paid to the evolution of the turbulent kinetic energy that was sampled by a combination of independent instruments. For the first time, this variable, which is a central variable in the turbulence scheme used in AROME and ARPEGE, is evaluated with observations.