After a long gestation period, JAGS 4.0.0 was finally released last week. If you go to the project page on Sourceforge then you should see an appropriate download link for your platform (binary packages for Windows and Mac OS X; source tarball for other platforms). Binary packages are also available for some Linux distributions. See the JAGS homepage for details.
Mac users should note that you need OS X 10.9 or later (i.e. Mavericks, Yosemite, or El Capitan). Older releases are no longer supported.
The rjags package for R has been updated to work with the new release of JAGS. It is not yet uploaded to CRAN, and the version of rjags that is available on CRAN (rjags_3-15) does not work with JAGS 4.0.0. However, you can download rjags_4-3 from Sourceforge. Again, binary packages are available for Windows (.zip) and Mac OS X (.tgz).
JAGS News 
Hi Martyn,
Do you have any plans for implementing the BUGS dloglik distribution?
Here are a few more details as to why I would like this feature. I recently developed a MCMC sampling framework with OpenBUGS+MATLAB (https://gitlab.com/reubenajohnston/mcmcBayes). Troubleshooting OpenBUGS sampler trap conditions has been frustrating me, so I ported mcmcBayes to support JAGS. Unfortunately, all of my models presently use the dloglik distribution, so the JAGS port isn’t complete yet.
Thanks,
Reuben Johnston
Following up to my previous note above…
I was able to get most of the model ports working using the infamous zeros trick. To my delight, JAGS performs significantly faster in comparison to OpenBUGS!
Regards,
REuben