This is confusing for most of us, no question on that. But again, loudness is an overall averaged thing, in a rather complicated bit of math.
For those delivering for broadcast work, they have to both make sure there are absolutely no peaks above X dB, depending on the local or national laws and regulations. AND they have to comply with a separate loudness setting, again to comply with local practices.
But they are separate things to check.
Maybe @PaulMurphy can pop in. He's quite an audio wiz.
If there's a difference between peak levels and loudness units (LU) between two edits, it simply means one has more dynamic range and likely less compression.
LU is based on perceived loudness over time. Because human hearing perceives some frequencies as louder than others, LUFS doesn’t just average levels—it weighs them based on how we hear. This makes it normal to see similar LU readings with different peak levels.
It’s also important to understand that visually seeing meters “hovering” around a certain range doesn’t equal the actual average loudness. For example, two recordings might both peak between -9 and -12 dBFS, but if one spends more time closer to -12, it will result in a lower LUFS value, even though the range appears similar