Stata 17 offers many new features in Bayesian analysis:
In addition, the bayesmh command supports many parametric survival models, which can be specified within multiple equations to fit, for instance, joint longitudinal and survival models. bayesmh also supports several new multivariate normal prior distributions with specialized covariance matrices such as exchangeable and identity. You can now specify exchangeable covariance structures with bayes: mixed. And last, but not least, you can specify Gibbs sampling for random effects with multilevel models fit by using the bayes prefix or bayesmh.
Learn more in the Stata Bayesian Analysis Reference Manual.