I would go with 6 - the number of PHYSICAL cores per NUMA node.
As other's said, Cost Threshold for Parallelism is VERY important. 5 is a universally too-low number these days.
Also note that Linchi Shea and Adam Machanic have done testing on some systems to show that > NUMA MAXDOP can be more efficient in some cases. Test, test, test! 🙂
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service