The paper presents a method to predict the absolute stability of proteins using a generative model, specifically the Evolutionary Scale Model 'Inverse Folding' (ESM-IF). The authors benchmark their approach on a dataset of 265 proteins, ranging from 30 to 150 residues, and find a mean error of 1.5 kcal/mol and a correlation coefficient of 0.7 for the absolute stability. They demonstrate that ESM-IF can predict the stability of proteins with two-state folding mechanisms and assess the relative stability between different conformations. However, the model struggles with more complex systems, such as multi-domain proteins and proteins with non-two-state folding mechanisms. The study highlights the limitations of current methods and suggests future directions, including the need for more accurate predictions of conformational free energies. The approach is simple to use and is freely available online.The paper presents a method to predict the absolute stability of proteins using a generative model, specifically the Evolutionary Scale Model 'Inverse Folding' (ESM-IF). The authors benchmark their approach on a dataset of 265 proteins, ranging from 30 to 150 residues, and find a mean error of 1.5 kcal/mol and a correlation coefficient of 0.7 for the absolute stability. They demonstrate that ESM-IF can predict the stability of proteins with two-state folding mechanisms and assess the relative stability between different conformations. However, the model struggles with more complex systems, such as multi-domain proteins and proteins with non-two-state folding mechanisms. The study highlights the limitations of current methods and suggests future directions, including the need for more accurate predictions of conformational free energies. The approach is simple to use and is freely available online.