The authors present a numerical model to simulate flow-like landslides from unstable rock slopes over a broad range of velocities (five orders of magnitude), and apply their model for the back-calculation of two events in the landslide area of Brienz, Switzerland. This topic is highly relevant from both a scientific and a practical point of view, and perfectly suits to the scope of the journal. The preprint is generally well written, structured, and illustrated, and I would certainly like to see this work published. I suggest some minor revisions based on the comments below, which might help to further improve an already very good paper.

• There are some minor issues of language and formatting (e.g., in L176 and L197, the closing brackets after the equation numbers are missing).
• Section 3.1: past tense and present tense are mixed, there should be consistent use of one of them.
• Caption of Fig. 9: Allgaü Schist -> Allgäu Schist
• 7 and Fig. 10: in Fig. 7, the observed outline is included in all panes, in Fig. 10 only for the last time step – maybe better to make this consistent.
• L290: You should also include work by others than Hungr and colleagues.
• L290f: There have been attempts of slow-flow simulation models also in GIS environments: the r.avaflow software now has a function for slower movements: https://link.springer.com/article/10.1007/s10346-023-02146-z and https://egusphere.copernicus.org/preprints/2025/egusphere-2025-213/ Recently, a dynamic earth  flow model was published: https://www.sciencedirect.com/science/article/pii/S0013795225000559
• L396: I am not sure about the statement “The failures documented at this site represent two landslide classes rarely considered …”. I think landslides of such type are often considered by practitioners. Maybe better reformulate the statement to make clearer what you mean.
• L402: research contact --> research contract

This paper presents a detailed dynamic analysis of two flow-like landslides (Igl Rutsch and Insel) in Brienz/Brinzauls, Switzerland. It introduces a novel GPU-accelerated numerical model (Orin-3D) capable of simulating landslide velocities ranging from moderate (meters/day) to extremely rapid (>5 m/s). The research is innovative, methodologically sound, and well-supported by field data, historical records, and modern monitoring techniques. The study provides significant theoretical and practical contributions to landslide hazard assessment and risk prediction. The paper is well-structured, logically organized, and supported by appropriate figures and references. Overall, the manuscript is of high academic merit and merits publication. Several suggestions are listed below for reference.

1. The paper notes that pore pressure dynamics were not considered in the simulations, which could be significant for clay-rich landslides. A more detailed discussion of this assumption’s implications should be added.
2. The model does not fully capture the seasonal velocity variations observed in the Igl Rutsch landslide. The authors should explore incorporating time-dependent parameters (e.g., viscous coefficient δ) to better match field observations.
3. A sensitivity analysis of key parameters (e.g., δ and φ) would strengthen the model’s interpretability.

This study addresses an important and specialized type of landslide—large failures in clay-rich rocks that transition into flow-like movements. The authors present a depth-averaged SPH model with a novel rheology capable of simulating slow-moving landslides (velocities on the order of m/day) and conduct detailed back-analyses of two well-documented events at Brienz/Brinzauls, which exhibited emplacement velocities differing by five orders of magnitude. The topic is highly relevant, and the manuscript is well-structured and clearly written. I recommend minor revisions before publication in NHESS.

Specific Comments:

1. The Lgl Rutsch event was simulated using 4,000 SPH particles. Could the authors clarify the rationale for this choice and discuss whether the results are sensitive to particle resolution?
2. The adopted SPH model is depth-averaged, which simplifies the real 3D dynamics. Please discuss how this assumption might influence the model’s predictive accuracy, particularly for flow transitions and velocity distributions.
3. Table 1 reports best-fit friction angles of 23°–28° for different stages. However, field and experimental studies (e.g., rock avalanches, earthflows) typically suggest much higher friction coefficients. Note that the reviewer is mainly working on fast-moving geophysical flows (e.g., Kong et al., JGR, 2023, 10.1029/2022JF006870). Are these smaller values physically justified for clay-rich rocks?
4. For the Insel event (Section 4.2), were the same SPH parameters (fitted to Lgl Rutsch) applied? If so, this seems inconsistent given the five-order magnitude velocity difference between the two events. Please clarify whether material or rheological differences explain this discrepancy.