Then it's almost certainly a cell problem, with more than one cell(group) given the large difference in voltage. If it was a sudden change and not gradual*** then it could be physical damage causing broken interconnects between cells within a group or groups.
*** (almost always is gradual, rarely is there an ebike/scooter/etc pack out there that doesn't need to cycle / balance at end of charge, especially once it has been used a few dozen times).
Either way, to fix this kind of problem, the cells all have to be replaced with new matched cells (identical capacity, internal resistance and other characteristics...not similar, but *identical*...if they are only similar, you still have a pack that gets unbalanced more and more every usage unless you leave it on the charger every time long enough to stop cycling on/off. )
If that's not an option, replacing the whole pack with a new one that uses new matched cells will also fix it.
If tha'ts not an option, but you want to "fix" this pack, you can replace just the cells (groups) that are filling up first, if they all started from the same low voltage, or if the pack starts out balanced at full charge then the cells that empty out first.
(cells that fill first, if all start from teh same "empty" voltage, have less capacity. Same for cells that empty first from the same "full" voltage).
BTW, 3p is not much for current delivery capability for most cells, and depending on your controller's current limit and how hard you push the system, can be pretty hard on the cells, causing them to age faster.
But 300-500 cycles for a lot of cells is a "good' lifespan, before they start getting below 80% of their capabilities, depending on usage. You'd have to look up the spec sheet and data from various testing sites for your specific cell model to see wha'ts realistic for those.
