FrontierScience: Evaluating AI's Ability to Perform Expert-Level Scientific Tasks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Miles, Lin, Robi, Hu, Kat, Jiao, Joy, Chowdhury, Neil, Chang, Ethan, Patwardhan, Tejal
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910004359987200
author Wang, Miles
Lin, Robi
Hu, Kat
Jiao, Joy
Chowdhury, Neil
Chang, Ethan
Patwardhan, Tejal
author_facet Wang, Miles
Lin, Robi
Hu, Kat
Jiao, Joy
Chowdhury, Neil
Chang, Ethan
Patwardhan, Tejal
contents We introduce FrontierScience, a benchmark evaluating expert-level scientific reasoning in frontier language models. Recent model progress has nearly saturated existing science benchmarks, which often rely on multiple-choice knowledge questions or already published information. FrontierScience addresses this gap through two complementary tracks: (1) Olympiad, consisting of international olympiad problems at the level of IPhO, IChO, and IBO, and (2) Research, consisting of PhD-level, open-ended problems representative of sub-tasks in scientific research. FrontierScience contains several hundred questions (including 160 in the open-sourced gold set) covering subfields across physics, chemistry, and biology, from quantum electrodynamics to synthetic organic chemistry. All Olympiad problems are originally produced by international Olympiad medalists and national team coaches to ensure standards of difficulty, originality, and factuality. All Research problems are research sub-tasks written and verified by PhD scientists (doctoral candidates, postdoctoral researchers, or professors). For Research, we introduce a granular rubric-based evaluation framework to assess model capabilities throughout the process of solving a research task, rather than judging only a standalone final answer.
format Preprint
id arxiv_https___arxiv_org_abs_2601_21165
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FrontierScience: Evaluating AI's Ability to Perform Expert-Level Scientific Tasks
Wang, Miles
Lin, Robi
Hu, Kat
Jiao, Joy
Chowdhury, Neil
Chang, Ethan
Patwardhan, Tejal
Artificial Intelligence
Computers and Society
Machine Learning
We introduce FrontierScience, a benchmark evaluating expert-level scientific reasoning in frontier language models. Recent model progress has nearly saturated existing science benchmarks, which often rely on multiple-choice knowledge questions or already published information. FrontierScience addresses this gap through two complementary tracks: (1) Olympiad, consisting of international olympiad problems at the level of IPhO, IChO, and IBO, and (2) Research, consisting of PhD-level, open-ended problems representative of sub-tasks in scientific research. FrontierScience contains several hundred questions (including 160 in the open-sourced gold set) covering subfields across physics, chemistry, and biology, from quantum electrodynamics to synthetic organic chemistry. All Olympiad problems are originally produced by international Olympiad medalists and national team coaches to ensure standards of difficulty, originality, and factuality. All Research problems are research sub-tasks written and verified by PhD scientists (doctoral candidates, postdoctoral researchers, or professors). For Research, we introduce a granular rubric-based evaluation framework to assess model capabilities throughout the process of solving a research task, rather than judging only a standalone final answer.
title FrontierScience: Evaluating AI's Ability to Perform Expert-Level Scientific Tasks
topic Artificial Intelligence
Computers and Society
Machine Learning
url https://arxiv.org/abs/2601.21165