We have a simulation with 100-1000 agents. When the number of agents gets larger, they sometimes pause, sometime for a long time (5-10 sec) when they are asked to make a new path. Even for very short paths. We run the simulation at higher timescale most of the time (2,4,8 times speed). The re-pathing pause issues are worse at higher timescales. We have some non-NavMesh pathing we do (with vehicles) that are unstable at higher speed, so we the simulation multiple times per frame to accomodate (so at 4x speed, we'll do 4 steps per frame) and this works really well to keep our simulation stable. Would there be any way we could make the NavMeshAgent do multiple steps per frame? Or is there another way to approach this issue? TIA, Peter.