I think this question comes from a misconception.

Rome used auxiliaries consistently throughout the Republican and Imperial period. Many have attributed "the fall of Rome" to reliance on auxiliaries. In Roman legend, the use of auxiliaries goes back at least as far as the reign of Tullius Hostilius, who supposedly destroyed Alba Longa for not providing the military support required. The army used non-citizens from the beginning, and didn't stop for some 700 years.

After 212 CE, by an edict of Caracalla, the entire population of the empire was made legally citizens.

They didn't.

Rather, "Roman Citizen" ceased to be a very meaningful distinction after Caracalla granted citizenship to every single inhabitant of the empire. With universal citizenship came the loss of special legal status, greater tax burdens, and eligibility to serve in the legions.

Still, even in much older times the "citizen" rule wasn't always strictly adhered to. I took part in a discussion on this sub a while back about whether or not Ceasar was recruiting non-citizens in his campaigns in Gaul. (Conclusion was: he may have, but not in large numbers. My source was overstating things.)

And after Cannae, the Romans in their desperation even asked slaves to join the legions, promising them freedom and citizenship upon discharge. They raised 8000 men this way. But clearly, this only happened in exceptional circumstances.

In the empire of late antiquity, meanwhile, the legion itself ceased to be the the primary building block of the army. In the Notitia Dignitatum we have a plethora of units and divisions. Some still go by the old names of cohorts, alae and legiones (though these appear to have been smaller in numbers: around 1000 men instead of 4-6K as in the old days) but terms like auxilia and vexilationes now appear to be used for high-prestige units, often with qualifiers like comitatenses (companions, of the emperor, i.e. part of the field army) and palatina (of the palace) as further indicators of status. Quite different from what they used to mean.

Whether the units labelled "legions" still consisted exclusively of citizenship at this point? I, quite frankly, have no idea. Perhaps. Diocletian made soldiering hereditary (If your father was a soldier, you had to be a soldier) and recruitment of citizens never stopped. Likewise, defeated enemies still were made to contribute men to the legions, just as they had in the early principate. But whether the distinction between recruits from inside and from outside the empire was maintained? Again, I just do not know. If anyone else knows more about the late Roman army, I'd be very interested to hear.