TurnBench
Leaderboard

Vox Maru v1

Submitted September 6, 2026 by Vox.

Read write-up

Test Performance

End-of-turn#1of 21
Recall
0.960
FP Rate
0.072
Latency (median)548 ms
Interruption#3of 19
Recall
0.964
FP Rate
0.085
Latency (median)368 ms

Rank is based on highest recall on the held-out test leaderboard.


Fetching and scoring the committed dev predictions…