JAGS is superb. It takes the BUGS language to another level. Before I saw JAGS I considered implementing my models by hand. Now I can't think of a reason: hardly any restrictions while most you could wish for is readily available.
>Flexible (far more than frequentist methods like lmer or glmer) >Excellent integration with R/rjags (much better than WinBUGS OR OpenBUGS) >Informative error messages (again, much better than *BUGS) >Active development (OpenBUGS appears stalled...) >64-bit builds allow fitting of large models >Helpful community
JAGS = flexibility of BUGS + seamless integration with R (with the help of rjags)
JAGS is a wonderful program for doing Bayesian models, especially if you are using Mac or Linux, I look forward to future development.
JAGS is just too awesome!