I'm having sporadic problems running code from the MDI window.  Sometimes I can run a few lines successfully several times and then it will just stop executing.  Sometimes a reset will get it working again.  Sometimes I have to restart the program to get it working again.  Sometimes I get an axis that will refuse to zero until I reset or restart.

I've not yet tried saving these command lines as a gcode program and running it that way to see if it executes correctly then.  That will be my next troubleshooting step.

Today, MDI refused to run a few lines giving me the error that line 0 did not have a real number.  Line 0 was f1.  I deleted the whole MDI window and pasted it back in and it still did that.  Reset would not fix it.  I restarted Mach and pasted the code back in the MDI window and it still gave me that error.  Very frustrating.  I put in a g0x1 then g0x0 at the top and it executed those two lines and then gave me the no real number error at the f1 line and stopped.

I've also noticed that sometimes when first starting, it will not jog from the keyboard (yes, it is enabled).  After a restart it will work.

Also, there is some kind of issue with the Mach 4 screen blinking off then back on (refreshing?).  It has done this as many as three times in succession and then it will execute the command I entered before it blinked off the first time.

Also, why are my A & B axis zero buttons greyed out?

I'm guessing that the 'A' and 'B' axis are greyed out because you haven't configured them properly in the Mach4 Control section.
As for the MDI issues.. while I haven't had that exact problem, I did have an issue with Mach4 (I also use an ESS) simply stopping during running. It was caused by several issues, one being some other background applications that were using two serial comm ports. Closing the applications resolved the problem. So if you are running anything else I'd close those applications and see if that helps. I also don't use the keyboard so you might try disabling that and see if the problem still persists. 

The A & B axes are as configured as I know how to get them.  The motor scaling and behavior is exactly as it should be when jogging or programming moves for those axes.  I can look further into it, but I don't know what more there is.

I also had one time where Mach stopped in the middle of something.  Don't know what it was.  I never run anything else.  The computer on it is dedicated to the mill, and I have internet disabled.

I don't have a choice except to use the keyboard.  My wireless keyboard (which is my 'pendant') is my only controller except for the hard wired estop button.  I never had a problem with it with Mach 3.

Mach3 and Mach4 are two completely different animals and shouldn't even be compared to each other.
is your program short? are you using a relatively lot of ESS IO?
Does your program end with an M30 or an M99? 

I am using a minimum amount of I/O for a 5 axis mill.  Inputs are E-stop, XYZ limit switches and probe for a total of 5.  Outputs, PWM, start, stop, 5pulse, 5 direction, 1 enable (all 5 axes on one enable) for a total of 14.

The program was simple and short.  I had run it several times successfully when it acted up.  It merely drilled 3 holes at 120 degree spacing around a center hole.  Code was as follows starting at the center hole 0,0.  I start the spindle then run this code from the MDI window.

f1 (plunge speed for drilling)
g16 (polar mode)
g0x.425y0 (first hole location)
g1z-.4 (drill first hole)
g0z0 (retract)
g0x.425y120 (2nd hole location)
g1z-.4 (drill 2nd hole)
g0z0 (retract)
g0x.425y240 (3rd hole location)
g1z-.4 (drill 3rd hole)
g0z0 (retract)
m5 (stop spindle)

The error it started giving out of the blue after I had already run it a couple of times was 'no real number where a real number should be in line 0', or something to that effect.

As I said before, I put a g0x1, g0x0 at the top to see what happened and it moved the x axis to 1 and back, then gave me the error again and stopped.  The f1 should not have been line 0 anymore after adding the other two lines at the top, but what do I know?

What gets me is it did the same thing after I copied those lines to the clipboard, restarted mach and pasted them back into the MDI window.

I've not had a chance to get back out to the garage and try typing them all in from scratch after a startup.

It's these sporadic, unexplainable things like this (and this is not the only one) that have me concerned about Mach 4.  It has features I definitely like better than Mach 3, but this unpredictability has me concerned that I may not have done the right thing when I purchased the license.  Maybe if I rolled back to a more thoroughly tested version, I would be better off.

Oh, I forgot the g15 (exit polar mode) that I had in there just before the m5.

I had an issue with a very short program.. it would randomly stop on a gcode line and simply wait... no fault.. and the ESS would miss changing states of outputs but would set the status as On or Off based on the command, not the actual state of the IO device. There were no errors in either case. Took me awhile to find workarounds.

I wouldn't test this in the MDI window.. use the normal gcode window.. then see if the problem persists.. don't forget to place an M30 at the end. You may also need to start the spindle inside the program though it should work if you start it manually.