Blog post · 2025
AstaBench: Rigorous benchmarking of AI agents with a holistic scientific research suite
| URL | https://allenai.org/blog/astabench |
|---|---|
| Topics | ReplicHAI / Agent Evals / 01 Research replication & science |
| Tags | agent-eval |
Abstract
Measures: Science agent suite + leaderboard. Metric: Score, cost.
BibTeX
@misc{ai2_astabench_2025,
title = {{AstaBench}: {Rigorous} benchmarking of {AI} agents with a holistic scientific research suite},
url = {https://allenai.org/blog/astabench},
abstract = {Measures: Science agent suite + leaderboard. Metric: Score, cost.},
author = {{Ai2}},
month = aug,
year = {2025},
keywords = {agent-eval},
}