For this kind of discussion, for me, it helps to think about what accuracy means.
Does it pertain to certain ways of testing? Blind tests with obvious transitions? No transition indicators? A given tube amp going into a load box vs the FAS unit with a similar impedance curve loaded and the same IR? Is accuracy about EQ only? Does "feel" factor into it and the ability of the player to discern differences that way?
Both answers of "no, it's not accurate" and "yes, it's accurate" seem relatively meaningless to me without further clarification.
Usually there's more said, sure. People often clarify further. But I do remember being told, many times, a given unit is simply "accurate", in the past, by people with technical expertise.. that I'd imagine would be more discerning about these terms, either way. "It's accurate!". Well ok, and what does that mean? The monkeys arse is blue.. ok. And I'm guilty of using that word vaguely too for sure.
My own short answer: for me, the relevant "accuracy" that matters the most does involve some EQ, as well as the "feel" of the amp sim (which isn't to say EQ does not affect feel; massively often does). And I do not have access to fractal's own amps either -- but the differences are often smaller between instances of amp models than I used to think years ago.
Now I went through a phase of offloading gear a couple of years ago. That included some tube amps, digital units and many guitars.
At the time, I was using ax8 often for recording. I got FM3 and had Kemper. I felt that Kemper was able to do a pretty good job EQ matching my tube amps. It could fool me on some blind tests, especially without careful listening, even if differences were usually obvious when playing. What was played also made a difference; I knew that in certain spots on my guitar neck they'd pop out more.
On that end, there were things about my Kemper profiles that felt relatively weird.
Some low notes wouldn't quite bloom like my amps would, even considering differences between my amps. There seemed to be a Kemper blueprint all around. It was hard to miss when monitoring was fine, especially for feel. It's as if there was this more limited "skeleton" of a tube amp Kemper had, and EQ matching wouldn't sufficiently mask it.
How did fm3 do?
Well -- in some cases I could spot fairly obvious EQ differences in ABX blind tests. I could also feel them while playing, tube amps through load box, same IR as Kemper and FAS and unit. And fractal didn't have amp sims representative of my amps here at the time.
And yet... where Kemper seemed to fail, at least to some relative extend (I say "relative" to emphasize this), the FM3 SEEMED to deliver. Even if some of the EQ was a bit off compared to Kemper, which EQ matches as part of profiting, usually I was happier by the fm3 end result.
This is still a limited experience for sure. Don't take this as some word of god. I'm a damn cat.
But having the tube amps there as well as fm3 and Kemper did make think that fractal modelling is quite accurate in the tests I conducted, according to own standards for this. Meaning: it could fool me in ABX tests, using the same IR, especially after some deep tweaking and targeting EQ in some more challenging cases (no eq matching, usually sufficient to go by ear).
In terms of feel, the response felt closer to something quite elemental in the real amps right there than Kemper managed to do.
In short: I believe fractal modelling is fairly "accurate". I'm pretty sure it isn't perfect, evident by the constant updates where there's always something that's now suddenly better and accurater. I doubt that train is about to come to a halt any time soon.
And many people seem to have a similar experience. They compare to their amps and the results are pretty good -- the gist of the tone. Not having fractal's amp isn't absolutely required to get a fairly good view of this imo, at least assuming the amplifiers aren't quite massively different.
But considering you specifically don't have access to amps (perhaps) close enough to the amp sims... I'm not quite sure what the best way to determine how close the sims are would be. You could go by tests some people post, of course. But the feel factor will be hard to judge, and there aren't as many tests of this nature for fractal amp sims like there are for some other units.
It's just not something people using FAS gear do as much as some others, for example Kemper users. But there are some good tests for sure, even if people don't have fractal's exact amps.