You can't rely on tech specs for lenses. They often lie about apertures and focal lengths, especially for telephoto lenses. Here are two taken at random: RF 800mm f/5.6, actual 780mm, f/5.9; EF 400mm f/2.8 II, actual 392mm, f/2.9. The correct values are given in the patents but they round the focal lengths up and the f-numbers down on the website.
Yeah but it's a lie that is widely known by people "who needs to know", while for the general public is pretty irrelevant if a 28-105 is a 30-102 and f4 is actually f4.2
And a 400mm being 392mm, in percentage, is even less relevant then with shorter focals.
Also, speaking about f stop, you're very well aware of difference between F stops and T stops; two nominal f2 lenses (meaning the focal length divided by the entrance pupil, let's say two different 100mm with an entrance pupil of 50mm) at same iso may require different shutter values due to light loss in transmission differences (different glass, different schemes, etc), so you may have, to bring the example to the limit, a f1.8 lens with a great transmission which is "faster", in terms of light captured (so faster shutter and/or less iso required), then a f1.4 lens with bad transmission (of course DoF is still different, shallower on the f1.4 lens and maybe with better bokeh rendering, but that's a different point).
So yes, lens data are all rounded, but still I don't find it really a relevant lie, as it's a widely available knowledge; like I won't find a real lie in "this camera reaches 40fps" and then you find out by testing that can only reach 38fps, that would be annoying, but still. Even "this thing weights 700g" and then it's 710g (or 690g), who cares, it's not payload to Mars were every gram counts.
Of course we should take everything with a grain of salt; but if I read anywhere "R5 II will be 60mpx, stacked, improved IBIS up to 10 stops, 80fps in electronic shutter, 8k240 video, battery last 5000 shots or 4h of 8k video, AI will recognize bride from groom at a wedding and will stay on him/her without AF switching and also recognize bride's dress and will control highlights on it reducing the current directly to the pixels covering the dress so it doesn't clip", well I want to read it black on white on Canon website before believing it, and I would probably just believe the first part (10 stops IBIS excluded), and totally doubt the second until I try it personally in the field and see with my own eyes the dress not clipping while the face remains of a higher luminosity.