You might submit conferences as a counterexample, but I would submit them in favor of my argument.
A preferable alternative in my mind would be a classroom where ALL work is done independently. The teacher then calls each student to their desk individually to see what they have learned and what they need help with. They can then spend 10-15 minutes individually with each student per day. Fast students can get pushed ahead and slow students can get the help they need instead of being forced along with the rest of the class.
So for one it takes 6 months to learn something, for another - 3 years. And in such systems it's fine - you got promoted because you have learnt something, not just because you got one year older (and haven't failed utterly).
And again, for the same "level" of subject there can be two speeds - so no risk of smart 8 yo kids in the same class with slow 11 yo.
Ability based learning within age groups makes perfect sense, but assumes a sample size large enough to make meaningful groupings at the high and low ends.