This study compares model simulations with metero radar (MR) observations. it presents the first systematic investigation of interhemispheric differences of mean winds and atmospheric tides at conjugate latitudes from observations and comprehensive models applying a unified diagnostic. It is a much needed study. While I appreciate the heroic effort of the authors, I am troubled by the lack of similarity between the model simulations described here and the MR observations. Previous studies - as properly discussed in this manuscript - revealed a much closer match between MR data and NAVGEM-HA. I am troubled by how badly the models are doing here.

The authors discuss the different behavior of the gravity wave drag parameterization in these models, arguing that the mean circulation is different to partialy explain the differences, i believe. I don't disagree in general, however such discrepancies seems to point to a fundamental flaw in these models: the lack of observations at MLT altitudes. Isn't that the take-home-message of this study?

Moreover, some of the models use atmospheric specifications (like MERRA): what is the time cadence of these atmospheric specifications? Typically these data are provided 6-hourly, which would not resolve semdiurnal variability, and in such case the comparison is between observations and the model's own climatology. In the same spirit, what it the nudging time scale? A long time scale would prevent the model to be tightly associated with the atmospheric analysis.

Specific comments:

Page 3, bottom: Why McCormack et al. (2015). That's a QBO paper. I think you want to use McCormack et al. (2017) as in the rest of the manuscript.

Figure 3 and similar figures. The authors really need to add contours: the color palette has a very large dynamic range and for the reader it is impossible to discern contours, especially when the largerst values are at the edges of the panels and the vast majority of the figure is a bland uniform color.  Also, I think would it help explaining the figures (and for us the readers, understanding it) if one hemisphere (say the SH) is rotate by 6 months, so that the same season is always in the middle.

Page 10, middle. Why is it expected that conjugate latitudes see almost the same behavior? 

The paper is a solid and well-founded comparison between northern and southern hemispheric MLT observations on the one hand and the comparison of the model output of three whole atmosphere GMCs with the observations on the other hand. The structure of the paper is logical and clear, but resembles more a technical report than a scientific paper.

The main presentation of the results is a mixture of contrasting observations from the two hemispheres for different locations and the same presentation for the model results (10 of 15 figures do this plus the five in the Appendix). As the technical methodology appears to be very sound, I only have two main remarks that might help modify the current manuscript and improve the presentation:

(a) I would suggest separating the physical comparison between the hemispherical observations from the comparision GCMs to the observations. This would allow the authors to formulate research questions that can be addressed and answered by the comparision. The comparision between the GCMs and the observations shouls constitute a second main part of the paper. Currently, it is a hudgepodge, hard to read and difficult to separate the individual results.

(b) The authors go to great lengths to create a homogeneous data set consisting of both observations and model outputs on comparable altitude-time grids. I wonder why the results are presented and discussed only qualitatively (" ... agrees reasonably well ...", etc). Why don't the authors show differences of the climatological means MODEL vs. OBSERVATION? I admit that the authors use a lot of effort to turn the shocking disagreements into positive words (e.g. " ....  shows a better agreement with the radars for the hemispheric zonal summer wind reversal ..." ) but for scientific usage a QUANTIFICATION of the differences would be really desirable!

Nevertheless, the study has its merits but really needs focus.