Establishing Best Practices for Building Rigorous Agentic Benchmarks
This addresses the need for reliable evaluation methods for AI agents, benefiting researchers and practitioners, but it is incremental as it builds on existing best practices and reported issues.
The paper tackles the problem of flawed agentic benchmarks in AI, which can misestimate agent performance by up to 100%, by introducing the Agentic Benchmark Checklist (ABC) to improve rigor, reducing performance overestimation by 33% in a case study.
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%.