I believe that the disagreement here is the term 'ampacity' means something different from 'maximum load the circuit can serve'.
Ampacity is a feature of the wire itself. It is the amount of current that the wire can carry on a continuous basis without overheating. The ampacity of #4 THHN is 95A. However because we have 4 current carrying conductors (4 conductors serving non-linear loads) we are required to 'derate' the ampacity of the #4 THHN to 76A.
I agree with Charlie: the 'final derated ampacity' of the conductors is 76A.
But if you ask about the maximum current that a circuit can serve, a separate '80%' factor comes into play that you must apply.
Normal breakers are limited to supplying no more than 80% of their trip rating when supplying continuous loads. This means, for example, that an 80A breaker is limited to supplying no more than 64A on a continuous basis, _unless_ you have a '100% rated breaker'. On top of this, the breaker is required to protect the wire.
These requirements combine to reduce the maximum current that a circuit can serve. For example, if you have conductors with 80A ampacity, they must be protected by an 80A breaker, and that 80A breaker means that the circuit can only serve a 64A continuous load.
Now there is a hole in my reasoning: since I've made the claim that the first 80% factor is the ampacity change created by the number of current carrying conductors, and the second 80% factor is forced by a breaker limitation, it would seem that I could have conductors of 76A ampacity, protected by an 80A breaker (round up rule), and serve a continuous load of 64A. However if you observe the requirements of 210.19(A)(1) the maximum load allowed to be served is 60.8A unless you have a 100% rated breaker, in which case the maximum load allowed to be served is 76A.
-Jon