This study compares the performance of the Community Land Models (CLM4.5 and CLM5) against tower and ground measurements from a tropical montane rainforest in Costa Rica. The study site receives over 4,000 mm of mean annual precipitation and has high daily levels of relative humidity. The measurement tower is equipped with eddy-covariance and vertical profile systems able to measure various micrometeorological variables, particularly in wet and complex terrain. In this work, results from point-scale simulation for both CLM4.5 and its updated version (CLM5) are compared to observed canopy flux and micro-meteorological data. Both models failed to capture the effects of frequent rainfall events and mountainous topography on the variables of interest (temperatures, leaf wetness, and fluxes). Overall, CLM5 alleviates some errors in CLM4.5 but CLM5 still cannot precisely simulate a number of canopy processes for this forest. Soil, air, and canopy temperatures, as well as leaf wetness, remain too sensitive to incoming solar radiation rates despite updates to the model. As a result, daytime vapor flux and carbon flux are overestimated, and modeled temperature differences between day and night are higher than those observed. Slope effects appear in the measured average diurnal variations of surface albedo and carbon flux, but CLM5 cannot simulate these features. This study suggests that both CLM models still require further improvements concerning energy partitioning processes, such as leaf wetness process, photosynthesis model, and aerodynamic resistance model for wet and mountainous regions.