Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wei, Jason, Sun, Zhiqing, Papay, Spencer, McKinney, Scott, Han, Jeffrey, Fulford, Isa, Chung, Hyung Won, Passos, Alex Tachard, Fedus, William, Glaese, Amelia
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2504.12516
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866916693604827136
author Wei, Jason
Sun, Zhiqing
Papay, Spencer
McKinney, Scott
Han, Jeffrey
Fulford, Isa
Chung, Hyung Won
Passos, Alex Tachard
Fedus, William
Glaese, Amelia
author_facet Wei, Jason
Sun, Zhiqing
Papay, Spencer
McKinney, Scott
Han, Jeffrey
Fulford, Isa
Chung, Hyung Won
Passos, Alex Tachard
Fedus, William
Glaese, Amelia
contents We present BrowseComp, a simple yet challenging benchmark for measuring the ability for agents to browse the web. BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information. Despite the difficulty of the questions, BrowseComp is simple and easy-to-use, as predicted answers are short and easily verifiable against reference answers. BrowseComp for browsing agents can be seen as analogous to how programming competitions are an incomplete but useful benchmark for coding agents. While BrowseComp sidesteps challenges of a true user query distribution, like generating long answers or resolving ambiguity, it measures the important core capability of exercising persistence and creativity in finding information. BrowseComp can be found at https://github.com/openai/simple-evals.
format Preprint
id arxiv_https___arxiv_org_abs_2504_12516
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents
Wei, Jason
Sun, Zhiqing
Papay, Spencer
McKinney, Scott
Han, Jeffrey
Fulford, Isa
Chung, Hyung Won
Passos, Alex Tachard
Fedus, William
Glaese, Amelia
Computation and Language
We present BrowseComp, a simple yet challenging benchmark for measuring the ability for agents to browse the web. BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information. Despite the difficulty of the questions, BrowseComp is simple and easy-to-use, as predicted answers are short and easily verifiable against reference answers. BrowseComp for browsing agents can be seen as analogous to how programming competitions are an incomplete but useful benchmark for coding agents. While BrowseComp sidesteps challenges of a true user query distribution, like generating long answers or resolving ambiguity, it measures the important core capability of exercising persistence and creativity in finding information. BrowseComp can be found at https://github.com/openai/simple-evals.
title BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents
topic Computation and Language
url https://arxiv.org/abs/2504.12516