cx_Freeze is still very effective for distribution - I use it all the time to produce a standalone version of my software and then I use Inno Setup to create an installer package for my customers.
As Loic states, Nuitka works differently from cx_Freeze in that Nuitka takes your Python source, compiles that to C++, then compiles the C++, whereas cx_Freeeze creates Python bytecode which is subsequently run by the included Python interpreter.
The result should be that the same Python code should run much faster in the form created by Nuitka, than the result created by cx_Freeze.
As Loic states, Nuitka works differently from cx_Freeze in that Nuitka takes your Python source, compiles that to C++, then compiles the C++, whereas cx_Freeeze creates Python bytecode which is subsequently run by the included Python interpreter.
The result should be that the same Python code should run much faster in the form created by Nuitka, than the result created by cx_Freeze.