One known bias in current Earth system models (ESMs) is the underestimation of global mean soil carbon (C) transit time (τ soil ), which quantifies the mean age of the C atoms at the time they leave the soil. However, it remains unclear where such underestimations are located globally. Here, we constructed a global database of measured τ soil across 187 sites to evaluated results from twelve ESMs. The observations showed that the estimated τ soil was dramatically shorter from the soil incubations studies in the laboratory environment (median as 4 with the interquartile range of 1–25 years) than that derived from field in-situ measurements (31 with 5–84 years) with the shifts of stable isotopic C ( 13 C) or the stock-over-flux approach. In comparison with the field observations, the multi-model ensemble simulated a shorter median (19 years) and a smaller spatial variation (interquartile range of 6–28 years) of τ soil across the same site locations. We then found a significant and negative linear correlation between the in-situ measured τ soil and mean annual air temperature, and the underestimations of modeled τ soil are mainly located in cold and dry biomes especially tundra and desert. Furthermore, we showed that one ESM (i.e., CESM) has improved its τ soil estimate by incorporation of the soil vertical profile. These findings indicate that the spatial variation of τsoil is a useful benchmark for ESMs, and we recommend more observation and modeling efforts on soil C dynamics in hydrothermal limited regions.