Establishing Best Practices for Building Rigorous Agentic Benchmarks

TL;DR

This paper identifies issues in current agentic benchmarks and introduces the ABC checklist to improve their rigor, significantly reducing performance overestimations in complex evaluations.

Contribution

The paper presents the Agentic Benchmark Checklist (ABC), a set of guidelines to improve the design and evaluation of agentic benchmarks in AI.

Findings

ABC reduces performance overestimation by 33% in CVE-Bench.

Many existing benchmarks have issues like insufficient test cases and miscounted successes.

Applying ABC improves the reliability of agent evaluations.

Abstract

Benchmarks are essential for quantitatively tracking progress in AI. As AI agents become increasingly capable, researchers and practitioners have introduced agentic benchmarks to evaluate agents on complex, real-world tasks. These benchmarks typically measure agent capabilities by evaluating task outcomes via specific reward designs. However, we show that many agentic benchmarks have issues in task setup or reward design. For example, SWE-bench Verified uses insufficient test cases, while TAU-bench counts empty responses as successful. Such issues can lead to under- or overestimation of agents' performance by up to 100% in relative terms. To make agentic evaluation rigorous, we introduce the Agentic Benchmark Checklist (ABC), a set of guidelines that we synthesized from our benchmark-building experience, a survey of best practices, and previously reported issues. When applied to CVE-Bench, a benchmark with a particularly complex evaluation design, ABC reduces the performance overestimation by 33%.