This is the critical question for current Stata 16 or 17 users.
For the first time, Stata 18 allows Bayesian estimation of multilevel models. This is a game-changer for education research (students nested in schools) or epidemiology (patients nested in hospitals). You can now fit: Stata 18
A major focus of Stata 18 is making "publication-quality" outputs easier to generate directly from the software: This is the critical question for current Stata