Antoines Georges has a nice article Thinking locally: Reflections on Dynamical Mean-Field Theory from a high-temperature/high-energy perspective.
The article is a short review dedicated to Dieter Vollhardt on his 60th birthday.
He discusses recent work which compares DMFT to high temperature series expansions for the Hubbard model. They agree for temperatures above about t, the hopping integral. This supports intuition that with increasing temperature physics becomes more local and so DMFT should be more reliable.
He also mentions cold atom experiments which are getting towards low enough temperatures that they could potentially be used to test DMFT predictions.