CMRR is an input referred parameter and shouldn't depend on circuit gain. For several reason it's not practical to measure CMRR in open loop:
Open loop gain is not specified and expected to vary with supply voltage, temperature and between exemplars. Also biasing in linear range is quite difficult.
Instead you prefer to measure CMRR in closed loop with well defined gain. Ratio of output voltage to differential input voltage variation can be easily calculated.
If you mean to test CMRR physically on test equipment then you need some auxillary hardware. There are old app notes about such methods which I have posted here before, but not stored on this phone.
In simulation you can put it in closed loop and "sniff" the input difference with a vcvs to get "open loop" numbers from two operating points. I have found stb unreliable without warning, so I use more primitive, inspectable setups.
Using Matlab to eval Aol and Fdbk ratio accuracy on CMRR quite revealing (the reason folks use
IA's which are laser trimmed in manufacturing to get performance) -