Tropical forests account for 29% of global soil carbon, but much of this belowground carbon is stored in pools with short (decadal) turnover times with a potential for rapid response to future change. Moisture may be more important than temperature in driving soil C storage and emissions in the tropics, yet the role of moisture on soil C dynamics is understudied and underrepresented in land surface models. We measured or attained data for soil carbon stocks and radiocarbon (¹⁴C) values of over 40 soil profiles from the Neotropics including sites in Puerto Rico, Mexico, Costa Rica, Panama, Brazil, and Peru. Our sites represent a large range of moisture, spanning 710 to 4200 mm of mean annual precipitation (MAP), and include Alfisols, Andisols, Inceptisols, Oxisols, and Ultisols. We compared measured soil C stocks and ¹⁴C profiles to data generated from global runs of the Community Land Model (CLM) v.4.5 and E3SM Land Model (ELM) v.1. We found a large range in soil ¹⁴C profiles between sites, and in some locations, we also found a large spatial variation within a site. MAP explains some of the variation in soil ¹⁴C profiles and carbon stocks, but differences in soil type contribute substantially to observed variation. We found that the models tended to overestimate carbon stocks compared to than measured stocks, while overestimating soil carbon age near the surface and underestimating the age of deep soil carbon. Additionally, the models did not capture the variation in ¹⁴C and C stock profiles observed in measured soil carbon profiles between and within the sites across the Neotropics. Ongoing and future work includes: 1) Expanded data for model-data comparisons with CLM5 and ELM, to better capture variation across the Neotropical region, and to evaluate the role of climate and soil properties to explain the variation in soil carbon age and stocks; 2) Collaborative site-specific studies to explore the influence of controlling factors in manipulation experiments and constrained gradients of precipitation, soil type, root inputs, geomorphology, and landuse on carbon storage and turnover; 3) Site-level runs of ELM v.1 and integration with a reduced complexity model (SoilR) will be used to evaluate model representation of soil C processes, including vertically-resolved carbon transfer rates, root inputs, and decomposition.