We use Simulation-Based Inference (aka, Likelihood-free inference) to constrain cosmological parameters with optical galaxy clusters. Using galaxy cluster observables (e.g., ) derived from the Quijote simulation suite, we train a machine learning algorithm to learn the joint probability distribution of the parameters that generated the simulations and the resulting galaxy cluster observables. This trained algorithm is then applied to a test set of galaxy cluster observables, to derive the corresponding cosmological and astrophysical parameters and their uncertainties. Our preliminary analysis shows that the posterior values of the parameters and their uncertainties are accurate when compared to the truth. These results demonstrate the potential of applying a Simulation-Based Inference to galaxy cluster cosmology studies.