I found this submission to be an interesting contribution to the field of coupled paleo-climate modeling. The study utilizes CLIMBER-X to effectively simulate the rapid ice growth over Eurasia and North America during this period and the ice decay after MIS 5d. The authors quantify the relative importance vegetation, ice sheet, and carbon cycle feedbacks play in the CLIMBER-X model for default parameters on ice growth and decay during the last glacial inception. They confirm the significant role dynamic vegetation plays in facilitating rapid ice growth, which is the most important of the tested feedbacks in their model simulations. Moreover, they confirm the importance of a temperature bias correction over North America for successful inception simulations with CLIMBER. However, their application of a summer bias correction throughout the year and constant through time must be explained. The bias correction enhances the agreement between simulated ice sheet configurations and geological records. The study's exploration of small temperature (+/- 1 degree C) and albedo (0.025) perturbations and their substantial influence on ice sheet volume and area adds further depth to the findings.

The content of this manuscript is relevant, shedding light on the complex dynamics of the last glacial inception and the factors influencing ice sheet growth and decay. The insights gained from this research with CLIMBER-X can contribute to the refinement of other paleo-climate models. However, the authors should comment on model and initialization uncertainty that can’t be addressed using a single simulation/model realization per experiment. 

I recommend this manuscript for publication with moderate edits. The authors have effectively addressed essential aspects of coupled paleo-climate modeling. With some adjustments, especially in explaining the assumptions made, this study will make a valuable addition to the body of literature in the field.

Major comments:

• The introduction doesn’t adequately prepare the reader for the results. What is actually novel in your study? While you present a literature review, you don’t show clearly enough where this manuscript fits in, which previous issues it addresses and what new knowledge it will contribute. The only time the present work is addressed is in the last sentence (“In this study we employ the Earth System model CLIMBER-X (Willeit et al., 2022, 2023) with interactive ice sheets, viscoelastic solid Earth response and dynamic vegetation to simulate the last glacial inception from 125 ka to 100 ka.”) which is not enough to guide the reader (who might not want to read the full paper but look for specific subsections) and stir interest. For example, lines 44-52 are unclear. Are you listing issues that previous studies had? Or important feedbacks that other studies have found that must be included to simulate the last glacial inception successfully? Here would be a good time to mention how your work will include/improve/explore said feedbacks and findings
• Limitations of this non-ensemble approach, uncertainties in parametric values and model initialization should be clearly stated
• Application of the temperature bias correction (around line 118). Why would you apply the summer bias correction throughout the year? I don’t see any reasonable explanation for that. What does the winter bias look like? And how can you assume the present-day bias is constant over time?

Minor comments:

• Line 23: Typo: Milanlkovitch
• Line 29: “relatively well covered by paleoclimate data”: is it well covered? What is “relatively”? Aren’t there significant uncertainties in any pre-LGM geological reconstructions?
• 84: “while Antarctica is prescribed at its present-day state in this study”: reasoning for this assumption?
• Line 99: “subsequently, temperature, humidity and radiation fields are downscaled onto the high-resolution topography.”: can you account for orographically forced precipitation on the high-resolution topography? 
• Line 101: “concentration of dust in snow”: what dust? Is there dust forcing? Are dust sources and transport simulated? Where the dust is coming from should be explained here in short and in more detail in the supplements.
• Lines 117-118: “we implemented a temperature bias correction over northern North America that has a dipole structure”: while explained in the supplements, it is not clear here if this is constructed or simply the JJA summer temperature field of ERA5 minus CLIMBER
• Line 199: Cite/compare to snowfield glaciation versus spreading from high-elevation nucleation sites in Bahadory et al. 2021: Last glacial inception trajectories for the Northern Hemisphere from coupled ice and climate modeling
• Table 1: unclear from the table what T offset, geo, and snow albedo offset are
• Figure 9 title: Zonal mean differences -> Northern hemisphere zonal mean differences
• Lines 232-236: structure: I’m missing a short experiment description before we dive into the results
• Line 254: “higher albedo of ice compared to ice-free land”: wouldn’t most of the now ice-free land be snow-covered? 
• Figure 14: why not also include the fixgeo experiment here?
• Figures 16, 17, 19: figure key consistently in the top panel like in other figures
• Line 321: “This result is fully consistent with the concept of glacial inception as a bifurcation in the climate system”: You haven’t really introduced the concept, and I don’t quite see how this plays a role here…
• Line 334: “A climate acceleration factor of 10 would allow more complex Earth system models to run transient glacial inception simulations in a reasonable time using less computational resources.”: Can we assume the finding still holds for more complex models? With the inclusion of more complex feedbacks and non-linearities, I would assume models can’t be accelerated as much
• Line 401: “A constant temperature lapse rate is used”: is assuming a constant lapse rate reasonable?
• Line 453: missing equation reference
• All ice sheet maps: the grayscale color key offers a poor discernable resolution

Potentially unnecessary figures if the paper needs to be shortened:  Figure 3, 18

Formal review of « Glacial inception through rapid ice area increase driven by albedo and vegetation feedbacks » by Willeit et al.

The manuscript presented by Willeit and co-workers is very much in line with other studies that have been conducted with similar complexity models at long time scales. As such, the aim is fine and the research presented is sound. Adequate attention is given to model parametrisation and process analysis, which is the strong point of the presented research. As presented however, the manuscript is an hybrid between a development paper and a research paper. It clearly lacks a research focus and is incremental on previous research with the same group. The latter point is not an issue and should not prevent publication.

Main concerns

1-/ Target journal and article format. The manuscript, as mentionned above is an hybrid between a development paper and a first research application paper. I find very surprising that the authors have chosen to pack all this in one manuscript, while other model developments manuscript are already published or in the way with the same first author in Geoscientific Model Development. Personally, I recommend that the current manuscript is splitted in two parts : a model development manuscript that would cover the themes of ice-sheet coupling, sea-level prediction model coupling and snow mass balance computation (including comparison to present-day fields since this is crucial for ice-sheet evolution). This would remove the large appendixes in the current manuscript and allow proper discussion of the modeling choices made by the expert community. A second, much lighter manuscript, would target the inception question with the model in a Climate of the Past research artcile. I see no good reason to proceed the way that the author did, yielding a manuscript that is complicated to evaluate correctly since other discussion are missing in an already too long manuscript.

Two examples of this issue :

line 114-120, page 5. Discussion of the constant bias correction over time in SEMIX is one example where this should be much more discussed and potentially be evaluated : we have many simulations in the different PMIP phases with GCMs that are performed for interglacials where you could test the validity of your stationarity of the bias (using other model anomalies and your own).

Line 100 : « fields are downscaled onto the high-resolution topography ». This is totally insufficient since most of the results of this manuscript and all the forthcoming with CLIMBER-X are dependent on the details of this downscaling. A detailed evaluation of the downscaling for rough topography should be given in a development manuscript. The current description of the equations in the appendix B1 is clear, but the absence of evaluation is unacceptable.

2-/ The main message of the manuscript is the relationship between surface albedo change (snow cover extent) and vegetation feedbacks that are at the source of the rapid ice-sheet expansion simulated, not the ice-volume. This is a fine conclusion, but is also one that has been largely promoted already by the same group (Ganopolski et al., 2010, doi :10.5194/cp-6-229-2010)

with the previous generation of their model. In the current version of the manuscript, little is said about the comparison to this previous results. Given the amount of components that are shared with the CLIMBER-2 model, it is in my view a requirement that such a comparison is made to assess what is really new in the study presented or at least what is mostly the same model response and what is not.

3-/ The discussion of the Figure 7 is not at an appropriate scientific level. Line 208 mentions that the model and data « compares reasonably well » which falls short of the mark. There is then a few consideration on the different places where the model is glaciated and not. However, there is in my view a fundamental problem in the ice evolution presented at 117 and 115 ka. Cited reconstructions seem to indicate that there is a double semi-independent ice-sheet build up, on one side on the Cordilleran ice-sheet and second over the northern part of Canada with more expansion over land in the Nunavut and Quebec areas. The model to the contrary indicates a very zonal expansion over the Hudson Bay, which is not what is indicated in the reconstructions (that are uncertain, but clearly indicate this more extensive expansion over Canada). The model also have a tendency to merge the two ice-sheets (Cordilleran and Nunavut/Northern Territories). What is the impact of all this on the results ? Likewise, very little is said about the clear expansion of the ice-sheet in Alaska (obvious tendency in figure 14 that is only partially corrected) and which was already a persistent feature of ice-sheet simulations in CLIMBER-2.

4-/ There is no discussion of the potential impact of a fixed Antarctic ice-sheet. It is mentionned at the beginning but then totally ignored. This is very much worrying since the authors are simulating sea-level. At the very least, a discussion of the potential impact, limitations etc. should be included.

5-/ In many places in the manuscript, the term « carbon cycle feedback » is used, but for me there is no carbon cycle feedback simulation in this research. The pCO2 of the atmosphere is fixed to reconstructions and vegetation is simulated on land (so probably the carbon as well) but not discussed. If the authors means « vegetation feedbacks » which is my guess there, then it should be corrected accordingly.

Minor concerns

1-/ line 22, page 2 « interglacial (no significant ice sheets over the northern continents) ». Please reformulate. I have a hard time not finding the current Greenland ice-sheet not significant.

2-/ line 236, page 14, « deciduos » → « deciduous »

3-/ line 172 : « continuosly » → « continuously »

4-/ line 181-182 : justification of such a starting point for the ice-sheet model is not justified. Why equilibrium at 125ka and not another condition ? How is this impacting the dynamics of the first part of your experiment, not having an transient evolution at the start ?

5-/ line 184 : another instance of « carbon cycle feedback », misleading