We employ a Bayesian method to infer stellar parameters from the PARSEC  v1.2S library of stellar evolution models and test the accuracy of these theoretical predictions. Detached eclipsing binaries are ideal for testing. We employ a compilation of 165 detached eclipsing binary systems of our galaxy and the Magellanic clouds with reliable metallicities and measurements for the mass and radius to 2 per cent precision for most of them. We complement the analysis with 107 stars that are closer than 300 pc, for which we adopted solar metallicity. The applied Bayesian analysis relies on a prior for the initial mass function and flat priors for age and metallicity, and it takes on input the effective temperature, radius, and metallicity, and their uncertainties, returning theoretical predictions for other stellar parameters of the binaries. Our research is mainly based on the comparison of dynamical masses with the theoretical predictions for the selected binary systems. We determine the precision of the models. Also, we derive distances for the binaries, which are compared with trigonometric parallaxes whenever possible. We discuss the effects of evolution and the challenges associated with the determination of theoretical stellar ages.