CentaurEval: Benchmarking Human-in-the-Loop Value in Agentic Coding

TL;DR

CentaurEval introduces a novel benchmark for evaluating human-in-the-loop value in coding, emphasizing collaboration between humans and AI to solve complex tasks that are intractable for either alone.

Contribution

It presents a new ecologically valid benchmark with collaboration-necessary problems, enabling standardized assessment of human-AI teamwork in coding tasks.

Findings

Collaboration significantly improves success rates from below 20% to over 30%.

Neither humans nor LLMs alone perform well on collaboration-necessary tasks.

Emerging co-reasoning partnership challenges traditional human-tool hierarchies.

Abstract

LLM-powered coding agents are reshaping the development paradigm. However, existing evaluation systems, neither traditional tests for humans nor benchmarks for LLMs, fail to capture this shift, excluding problems that require both human reasoning to guide solutions and AI efficiency for implementation. We introduce CentaurEval, a unified, ecologically valid benchmark for measuring human-in-the-loop value in coding. CentaurEval's core innovation is its "Collaboration-Necessary" problem templates, which are intractable for standalone LLMs or humans, but solvable through effective collaboration. CentaurEval dynamically instantiates tasks from 45 templates, providing a standardized IDE for humans and a reproducible 450-task toolkit for LLMs. We benchmark 45 participants against 5 LLMs under 4 levels of human intervention. Results show that while LLMs or humans alone achieve poor pass rates (0.67% and 18.89%), human-AI collaboration significantly improves to 31.11%. Our analysis reveals an emerging co-reasoning partnership, challenging the traditional human-tool hierarchy by showing that strategic breakthroughs can originate from either humans or AI.