The one with the impedance curves and the one with the gain enhancer thing were big for me. But I have been largely happy since the Axe II days, so I do not get blown away every time an updated FW is pushed. Always love new models and FX but not gonna be 'WOW' over them.
Always appreciate the effort and continued support, though, some good things happen incrementally.
Guys, imagine he wanted to do an "Evolution of the Axe FX modelling" video, where he showed how the sounds improved over different versions. It'd be nice to know which updates made noticeable sonic improvements. "The latest is the best" is nice and all but doesn't help show the evolution.
You'd have to reset/redial models when you cross a major FW boundary, so the question would always be did you actually recover from that properly via a redial... there's no really good way to do that experiment, I don't think.
Honestly some of the best modeled tones I've ever heard were Mark Day's AxeFx Ultra tones on FW 9.x for example. Jst because he was getting great results to my ear... A one FW to another comparison with no point of reference is probably just not useful.
The only really valid way to measure progress of a simulation is to somehow do a data based comparison/measurement of the behavior of the physical hardware and your simulation; almost anything else is just wanking. I'm sure that Cliff uses his ears and judgement as part of the process but I am also sure he is constantly experimenting and advancing things using objective measurements. Otherwise you'd end up in the weeds.
Again, love FAS, and the updates, but the vast majority of FW updates are just 'different' rather than better for me on first blush. It has been like this for most updates for me, for a long time since FW12+ on the II. IMHO.