I think the LuaJIT maintainers have some issues with recent additions to the language, and with the direction that the Lua team has taken. So yes, frozen to 5.1 for the foreseeable future except for some features from more recent versions.
The issue linked in the sibling comment is the main one, but for a specific example of them having reservations about a feature in a more recent version, there's this:
That's an awesome example, thank you. I'm making my own language and it's hard to overstate how thorny the evaluator and call stack code can get once seemingly neat features are added. I'll try to avoid making the same mistakes. I agree with him that implicit scopes for resource management serve no purpose in a dynamic language.
I think we can concede that there is a legitimate language fork here. I personally was not all that happy with the extent of the changes in 5.3. I would be just as happy at 5.1.
