Abstract: Model evaluation is an indispensable part of model development. In this study, the authors evaluated the Common Land Model (CoLM) using the latest benchmark software called International Land Model Benchmarking (ILAMB) and compared this model with the Community Land Model version 5 (CLM5) from NCAR. As a land surface model benchmark software, ILAMB can automatically generate graphical diagnostics for model variables and score model performance. Results show that CoLM generally performs well, and its simulation result is close to the benchmark data. Compared with CLM5, CoLM is slightly inferior for gross primary productivity and hydrologic processes. For radiation, CoLM performs as well as CLM5, and is even better for variables such as surface upward long-wave radiation and surface net radiation. Comparing the forcing data of CRUNCEPv7 and GSWP3v1, the authors found differences in the climate average state and observed that the model performance under GSWP3v1 forcing is better. The latent heat flux simulated by CoLM and CLM5 under CRUCNEPv7 forcing has a significant positive bias in the Amazon plains, eastern and southern Asia, and eastern North America, but the bias decreases under GSWP3v1 forcing. The two models overestimate the sensible heat flux in north Africa and central Asia with either of the two forcing datasets. Regarding the radiative process, the surface upward shortwave radiation simulated by CoLM was mainly on the high side in the world, which to a certain extent caused the low level of surface net radiation. The surface upward long-wave radiation scores of four experiments are similar, but a difference exists in the specific spatial distribution.