The authors quantified the future ozone improvement from the Net Zero policies by using the chemistry-climate model and deep learning model, and examined the different drivers of such changes in designed scenarios. I think this paper is interesting, and this topic of great importance to policy makers. However, several major revisions should be addressed before the recommendation.

First, unlike the PM_2.5 pollution, ozone pollution is heavily affected by the meteorological variations from historical experiences (e.g., the period of 2013-2020), especially in the context of future warming climate. However, I didn’t see the clear descriptions of meteorological data used for this study in the Method part, and how it changes over times and its effect and the interaction with climate change on future ozone pollution.

Second, is there any limitation or uncertainty by further applying the deep learning model (i.e., the correction ratio of historical results) to correct the biases in surface O₃ simulated with UKESM1 in future analysis?

Third, the authors should reorganize the result part, I cannot understand that why the “Bias corrected surface O₃ under the Net Zero pathway” part is presented at last in the results part as it is the basement of ozone improvement and drivers analysis, I think.

Fourth, several limitations and uncertainties indeed exist in this study, for example, the model itself and meteorological data used, which should be systematically presented at the end of the manuscript.

Minor comments:

• Add quantitative results in the Abstract part.
• Add more introduction of current ozone pollution condition in the first paragraph.
• Please check and revise all the maps with 9-dash-lines in China.
• Please explain the effect of external emissions outside China (like a circle) in Figure 2b (probably wrong, please check).
• Part 3: more quantitative results (i.e., numbers) should be added.
• Figure 3, again, the small effect of climate on ozone pollution seems weird for me. Please add more explanation.
• As I know, generally model simulation has relatively bad performances on high ozone concentration (i.e., summer), while from Figures 6c and 6f, the model seems be not good in March, April (not the period of high values), please explain the reason.
• There are many recent studies analyzing future ozone pollution in the context of carbon neutrality, please add the comparison.

This manuscript simulated surface ozone changes in China under a Net Zero pathway using a chemistry-climate model which have corrected biases through deep learning model. The results indicated a substantial decrease in the monthly average surface ozone concentrations in the summer, with the greatest contribution from local emission reductions. The entire study appears technically sound, and the results are well interpreted. Thus, I recommend the publication by addressing the comments below.

1. For line 125, In addition to ozone, are there any other variables included as input in the deep learning model?
2. Why does methane contribute more significantly to ozone increase in western China compared to eastern China?
3. For figure 7d, the authors suggest that the significant smaller decreases of latitudinal mean surface O₃ implies underlying impacts of emission controls on O3 may not be as large as the model suggests, and the overestimation of O₃ responses to emission changes. However, by comparing figure 7 (a) and figure 1 (a), it can be observed that the current smaller decrease between the two scenarios after correcting with deep learning model is primarily due to a reduction in the corrected ozone concentrations. Does this imply that the decrease of ozone cannot fully demonstrate the reduction in pollution emissions?