Force spectroscopy techniques are often used to learn about the free energy landscape of single biomolecules, typically by recovering free energy quantities that, extrapolated to zero force, are compared to those measured in bulk experiments. However, it is not always clear how the information obtained from a mechanically perturbed system can be related to that obtained using other denaturants, since tensioned molecules unfold and refold along a reaction coordinate imposed by the force, which is unlikely meaningful in its absence.Here, we explore this dichotomy by investigating the unfolding landscape of a model protein, which is first unfolded mechanically through typical force spectroscopy-like protocols, and next thermally. When unfolded by non-equilibrium force extension and constant force protocols, we recover a simple two-barrier landscape, as the protein reaches the extended conformation through a metastable intermediate. Interestingly, foldingunfolding equilibrium simulations at low forces suggested a totally different scenario, where this metastable state plays little role in the unfolding mechanism, and the protein unfolds through two competing pathways 27 . Finally, we use Markov state models to describe the configurational space of the unperturbed protein close to the critical temperature. The thermal dynamics is well understood by a one-dimensional landscape along an appropriate reaction coordinate, however very different from the mechanical picture. In this sense, in our protein model the mechanical and thermal descriptions provide incompatible views of the folding/unfolding landscape of the system, and the estimated quantities to zero force result hard to interpret..