cd/entity/Long-Horizon-Terminal-Bench· home› entities› Long-Horizon-Terminal-Bench

grep -l @long-horizon-terminal-bench /news/*.json | wc -l → 1

Long-Horizon-Terminal-Bench

mentions 1 type Organization feed RSS

// recent coverage 1 mentions

04:00

2026-07-13

arxiv.org

artificial-intelligence

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

Researchers introduced Long-Horizon-Terminal-Bench, a benchmark of 46 long-horizon tasks designed to evaluate AI agents on complex, multi-step workflows with dense intermediate rewards. Testing 15 fro…

// co-occurs with top 1 entities

arXiv 1