← back to paper

Review history

arxiv: 2607.08964 · 2 revisions

Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading

1. 2026-07-14 CONDITIONAL HIGH v1.1.0-grok45 novelty 6.0
   41660 ms 23367 in 3660 out 2026-07-14T15:23:48.745165+00:00
2. 2026-07-13 CONDITIONAL HIGH v1.1.0-grok45 novelty 5.5
   44081 ms 22954 in 3517 out 2026-07-13T05:27:10.169857+00:00