Ran a sim to put things in perspective.
Solver wants to cbet range basically, allowing for a 33%/75% sizes it chooses the smaller one 80/20. K
T
bets 90% of the time for a 33% size, almost never checking.
OTT once villain calls it's a 50/50 from checking or actually potting (!), no 25% or 50% stuff, so solver wants to polarize.
K
T
is a mix, if checked vs villain's bet it mostly folds, even though there are some some x/r vs smaller sizes.