The authors reassess the apparent disagreement between observed Antarctic sea ice trends and CMIP models, in light of recent extreme low sea ice events and the impact on the observed trend field. This is a valuable exercise that gives important context to the reliability (or otherwise!) of coupled models for the Antarctic climate system. Unfortunately I think the execution needs to be improved before recommending publication:

1) the treatment and presentation of the literature on Antarctic SIA trends isn't really adequate, and this does have an impact on the interpretation of the results/discussion. It's not really clear from the opening paragraph whether the author's are claiming a discrepancy between modelled and obs trends to 2014, but the language implies that (e.g. "...consistent with observations."). That's not really true in the literature - at least not for total SIA, when the model internal variability is properly accounted for (zunz et al 2013, polvani and smith, 2013) - it's only when spatial trends were considered that the model trends were incompatible with obs (Hobbs et al 2015). I think the Intro needs a very clear statement from the authors about what they mean by agreement with obs (also in the Discussion section), and a bit more nuanced outline of the literature to-date

2) My biggest concern is that I don't think the handling of the CMIP6 data is adequate.

2a) Firstly there is no drift correction which is essential for dealing the historical simulation trends. Most models will have little drift in SIA but some on the list  (e.g. MIROC) are known to have quite large drifts. I think at the very least proof from the piControl experiments that spurious trends in SIA are small is required

2b) internal variability in the models isn't properly dealt with. The method used implicitly assumes that the models all have similar internal variabilities but this assumption isn't stated and isn't really valid - some models (e.g. GFDL) have some pretty large multidecadal internal variability (Zhang et al 2018) that differs greatly from other models. Even by truncating the max contribution of each model to 6 ensemble members, there's still a weighting towards those models with more members. As for the model drift correction, interrogating the piControl experiments is the correct way to represent modelled internal variability. Aside from internal variability, quite a few models have ice-free summers for the period of interest (Roach et al 2020) which is obviously going to impact their trends (no ice = no trend)

3) obs - I assume that the 'synthetic' extension of the obs record to end of 2023 is just a placeholder, and that the actual data will be used before publication? 

Hobbs, W. R., N. L. Bindoff, and M. N. Raphael, 2015: New Perspectives on Observed and Simulated Antarctic Sea Ice Extent Trends Using Optimal Fingerprinting Techniques. J Climate, 28, 1543-1560, 10.1175/JCLI-D-14-00367.1.

Polvani, L. M., and K. L. Smith, 2013: Can natural variability explain observed Antarctic sea ice trends? New modeling evidence from CMIP5Geophysical Research Letters, 40, 3195-3199, 10.1002/grl.50578.

Zhang, L., T. Delworth, W. Cooke, and X. Yang, 2018: Natural variability of Southern Ocean convection as a driver of observed climate trends. Nature Clim. Change, 10.1038/s41558-018-0350-3.

Zunz, V., H. Goosse, and F. Massonnet, 2013: How does internal variability influence the ability of CMIP5 models to reproduce the recent trend in Southern Ocean sea ice extent? The Cryosphere, 7, 451-468, 10.5194/tc-7-451-2013.

Holmes et al. compare observed Antarctic sea ice area trend with the trends in CMIP6 simulations. The authors find that during 1979-2018, the models are not consistent with the observations, as has been widely noted previously. This is because the Antarctic sea ice area tends to steadily decline in response to anthropogenic forcing in GCMs, whereas the observed Antarctic sea ice area increased during the first 35 years or so of the satellite record (roughly 1979-2015). However, strikingly low Antarctic sea ice areas were observed in 2022 and 2023. Due to this, the authors find that when the time period over which the trends are computed is extended to 1979-2023, the models and observations are fairly consistent. The authors conclude that this gives a greater level of confidence in the CMIP6 Antarctic sea ice simulations than previously though.

The manuscript is clearly and concisely written, and the presentation is polished.

However, I do not find the conclusions to be compelling. The issue is that although we often use linear trends to compare models with observations, the observed Antarctic sea ice area evolution looks strikingly unlike a linear trend plus any straightforward type of noise. This can be readily seen by looking at any time series of observed Antarctic sea ice area or extent (e.g., https://zacklabe.files.wordpress.com/2023/12/nsidc_sie_timeseries_ant_anomalies-5.png). I do not see any meaningful similarity between the steady long-term decline of Antarctic sea ice area in typical CMIP6 simulations (e.g., Historical and then SSP5-8.5 simulations plotted in Fig 4c,d of Roach et al. 2020) and the observed gradual expansion followed by a short-lived abrupt loss during the past two years, even if both time series have similar OLS linear trends. In my opinion, it is misleading to use the similarity between these linear trends to conclude that “we should now have some level of greater confidence” in the model simulations and that therefore “projections of substantial future Antarctic sea ice loss may be more reliable than previously thought” (quotes from the manuscript). I am not sure what meaningful information can be gleaned from noting this similarity in linear trends between the observations and GCMs during 1979-2023. To this end, the authors do not include any plots of actual time series, only reporting the OLS linear trends, so a reader unfamiliar with the observations and simulations may be substantially misled by this manuscript. (I don't mean to imply that the authors are being intentionally misleading in any way.)

Furthermore, the authors mention that the observations are consistent with “the ‘two-timescale’ response to stratospheric ozone forcing whereby increasing westerlies cause a sea ice increase on ‘short’ timescales and decline on ‘long’ timescales (Ferreira et al., 2015; Kostov et al., 2017).” It should be emphasized that a later study that included some of the same authors (Seviour et al. 2019, doi:10.1175/JCLI-D-19-0109.1) concluded that this ozone forcing mechanism is unlikely to be a primary driver of the mismatch between the Southern Ocean surface cooling (and Antarctic sea ice expansion) in observations and the Southern Ocean surface warming (and Antarctic sea ice retreat) simulated by GCMs.

Hobbs et al. compared trends of observed Antarctic sea ice with those of CMIP6 simulations. This investigation sheds light on the model's performance and its ability to capture and reflect the dynamic changes in Antarctic sea ice in the face of recent, dramatic declines.

The authors find that between 1979-2018, the model did not compare well with the sea-ice observations, since the SIE/SIA increased over the first 35-years and the model concurrently showed a decline. However, when the analysis was extended to include the 2022 and 2023 lows (encompassing the full satellite record) a more favorable alignment between the model and observations emerged. They concluded that this was potentially due to being able to assess the trend over a longer time period, which gives a greater confidence to CMIP6 simulations.

Overall, I am not 100 % convinced with the authors’ conclusions. The CMIP6 model has consistently underestimated the SIE and SIA in the Antarctic, and thus did not reproduce the 1979-2014 sea-ice trends. Although the model seems to align better with recent data showing a decline in sea ice trends, especially in 2022 and 2033, I find the authors' argument insufficient to establish the model's reliability and instill a higher level of confidence in its capabilities.

It would be beneficial for the authors to provide a more detailed account of the enhancements made to CMIP6 that contribute to its improved simulation of sea-ice trends, particularly when considering the extended satellite record, and recent sea-ice lows. Given the well-documented temporal and regional variations in Antarctic sea ice, where trends and regions often exhibit stark differences, a more comprehensive explanation of the model modifications would help readers better understand the factors leading to its purported improvement in capturing these complexities.

Specific comments:

Line 54: Figure B1 is referenced here and in two other places, but there is only a Table B2 in the Appendix B.

Lines 81-82: This sentence is a quite confusing and convoluted.

Line 105: I would suggest changing the sentence to “This makes it less likely that the observed...”