I did a quick sim of a similar circuit and yes it works

I used different devices/resistances etc and only the 2nd and Output stages. You get a little DC at the output (I had about 800mV) due to the 10K bias and 2K refrence resistors but, that's really not a problem...
I'll go play a little more... I'ld like to see just what's really going on here and what the current ballance really looks like. Of coase the class A, AB etc also...
Just as I thought. Although looking at the load voltage, and as I'm using a resistor, the load current, it looks class A. But Looking at the source currents in the output devices you are only seeing 1/2 the signal on each source. Class B...