Abstract. Assessing global models in the upper troposphere (UT) and in the lowermost stratosphere (LS) is an important step toward a better understanding of the chemical composition near the tropopause. For this purpose, the current study focuses on an evaluation of long-term simulations from four chemistry–climate/transport models, based on In-service Aircraft for a Global Observing System (IAGOS) measurements. Most simulations span the period from 1995 to 2017 and follow a common protocol among models. The assessment focuses on climatological averages of ozone (O3), water vapour (H2O), carbon monoxide (CO), and reactive nitrogen (NOy). In the extra-tropics, the models reproduce the seasonality of O3, H2O, and NOy in both the UT and LS, but none of them reproduce the CO springtime maximum in the UT. Tropospheric tracers (CO and H2O) tend to be underestimated in the UT, consistently with an overestimation of cross-tropopause exchanges. Most models systematically overestimate ozone in the UT, and the background of nitrogen oxides (NOx) appears to be the main contributor to ozone variability across the models. The partitioning between NOy species changes drastically across the models and acts as a source of uncertainty in the NOx mixing ratio and on the impact of these species on atmospheric composition. However, we highlight some well-reproduced geographical variations, such as the Intertropical Convergence Zone (ITCZ) seasonal shifts above Africa and the correlation of extratropical ozone (H2O) in the LS (UT) with the observations. These features are encouraging with respect to the simulated dynamics in both layers. The current study confirms the importance of separating the UT and the LS with a dynamical tracer for the evaluation of model results and for model intercomparisons.