DETOUR: An Interactive Benchmark for Dual-Agent Search and Reasoning

Li Siyan, Darshan Deshpande, Anand Kannappan, Rebecca Qian · 2026

Read DETOUR: An Interactive Benchmark for Dual-Agent Search and Reasoning by Li Siyan, Darshan Deshpande, Anand Kannappan, Rebecca Qian (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-00352.txt.

Description

Li Siyan and team introduce DETOUR, an interactive benchmark for dual agent search and reasoning. Recalling information in conversation often takes multiple turns, yet existing benchmarks for tip of the tongue search cover only single turn settings. A Primary Agent, the subject of evaluation, identifies the recollected entity by querying a Memory Agent held consistent across evaluations. Results show leading models still struggle, reaching only 36 percent accuracy across all modalities, text, image, audio, and video, underlining the need for stronger underspecified search abilities.

Catalog record

CREATORLi Siyan, Darshan Deshpande, Anand Kannappan, Rebecca Qian
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47448
LICENSEhttp://creativecommons.org/licenses/by-sa/4.0/
RIGHTSCC BY SA
SOURCEarXiv

Source: https://arxiv.org/abs/2602.00352

Browse the full library · About the Tower of Babel