Yes, that matches my experience so far.

Yes, that matches my experience so far. Here is another Jev versus Luna Medium example. Given the same input, Luna maneuvered more effectively (although at x5.4 the cost and x5.1 the time) and avoided pursuing the apple when doing so would create a dangerous position in the future. I understand this may be outside the system’s intended scope, but reliable long-horizon decision making at this speed would be incredibly useful. It can only improve from here.