CMU releases SearchHorizonBench for agent search decisions

CMU released SearchHorizonBench, a benchmark built on 45,632 executions that measures whether AI agents can decide if another web search is worth its cost.

2026-10-28 ~ 2026-10-28 · 2 related posts

• SearchHorizonBench: 45,632 runs to teach agents when another search is worth it — _reachsumit · 2026-10-28
• CMU's SearchHorizonBench Measures Whether Agents Know When More Search Pays Off — CarnegieMellonU · 2026-10-28