Tasks

Real Unreal Bench · Task 15

Play an athletics game against its spec and fix it

Task pass rate88% ±22%

Before an athletics game ships, the agent plays it against a full spec (the four-event line-up, the menu and its athlete, the meet, throwing and jumping, the run-up camera, hurdling and sound), finds where it falls short and fixes it, without breaking what works.

Before

A throw called a foul

After

The same throw measured and leading
The grader's frames of the same moment: the untouched start, and the reference solution.

Results