Constructing phylogenetic trees in BEAST package and MrBayes requires an AIC selected best-fit substitution model. Running a model search in mrmodeltest requires a script created in PAUP.
I think IQTree is the most comprehensive phylogenetic model tester currently. It is also easy to use and has a well-written manual.
If you want to test by Bayesian cross-validation this can be done under PhyloBayes. However, on reasonable sized matrices you'll need a lot of computation power, or a lot of time on your hands :)
If you want to go quick-n-dirty without model testing, just apply Le & Gascuel (LG), plus a gamma distribution on rate. For multicellular nuclear sequence data, LG is very often best fit. This will not be true for other types of data, particularly epidemiological data.