The Le Teil earthquake is a typical case of a moderate, high-impact earthquake. It is important to test risk calculation models for such earthquakes. The subject of the article is therefore important and corresponds to the interests of the readers of this journal and special issue. However, the paper needs to be significantly improved. The article is difficult to read and somewhat disappointing, given the expectations associated with the subject and the authors' considerable expertise. The major results are insufficiently identified and described, and I found it hard to understand which factors have the strongest impact on the damage estimates, as it is difficult to understand which parameters are fixed and which parameters change in the calculations presented. I agree with reviewer 2's recommendation. The article should focus on a sensitivity study of seismic motion prediction with a clear and rigorous methodology. Both reviewers give very useful recommendations. I therefore recommend that the manuscript (which has a strong potential) be revised to take these suggestions into account.

The paper is much improved and the results are better presented and discussed. One reviewer provides a solid list of recommendations/comments that should be considered in the final version.

I also have some suggestions/recommendations:

- The abstract should be strengthened to better reflect the lessons learned from this analysis. The current last sentence is weak and does not provide solid information to the reader. I suggest adding to the abstract the key "lessons learned" to improve future testing, which are listed in the conclusion.
- The authors compare many models, but do not provide enough information on which models are the most "solid/likely". Of course it is difficult to compare models, but the paper would really be more useful if the authors could give an expert opinion on them. For example, the paper states that the ESRM20 and brgm exp exposure models are different, but it is not really clear which of the two models is considered the most reliable (would the authors give similar weights to them in a logic tree approach?)
- It is not clear to me how low-cost sensors can contribute to a better explanation of damage levels (last sentence of the conclusion).
- The conclusion discusses the use of the AS2000 model (Atkinson and Sonley, 2000 ?). This model is not mentioned in the analysis or in the bibliography. The use of many acronyms AS2000, FM2010, Ko2020 makes the paper difficult to read (the link with the reference list becomes difficult). I suggest to stick to the classical way of citing papers (Atkinson and Sonley, 2000 instead of AS2000).
- Figure captions ("model") to the right of Figure 1 and 2 are not necessary.