多智能体真的更有效吗？评估 LLM 多智能体工作流的准确率与成本

View PDF HTML (experimental)

Abstract:Choosing whether to replace a single-agent controller requires evidence on both task performance and inference expenditure. We present BenchAgent, a shared evaluation framework, and compare six multi-agent workflow configurations against its single-agent reference, BenchAgent Core, across ten reasoning, coding, and tool-use benchmarks. All configurations use GPT-4.1 and common tool and scoring interfaces; Core executions share the same step limits. We measure accuracy and recorded token use over three runs on fixed task sets, with 3,302 instances per configuration per run. EvoAgent reaches 79.70% mean pooled accuracy versus 76.65% for Core and exceeds Core in all three runs; the other five MAS trail by 1.71-8.76 percentage points. LLM-Debate gains 9.67 points on HumanEval but loses 1.96 points in pooled accuracy while consuming 1.39 times the tokens of Core. ChatEval gains 22.25 points on IFEval but trails Core in pooled accuracy with 4.34 times its token consumption. The broad comparison and targeted analyses of retained responses provide evidence for workflow selection and identify answer preservation and resource allocation as directions for subsequent multi-agent optimization.

Comments:	27 pages, including supplementary material. Revised evaluation and results; updated author list. Code: this https URL
Subjects:	Artificial Intelligence (cs.AI)
Cite as:	arXiv:2606.05670 [cs.AI]
	(or arXiv:2606.05670v2 [cs.AI] for this version)
	https://doi.org/10.48550/arXiv.2606.05670 arXiv-issued DOI via DataCite

Submission history

From: Yuhang Fu [view email]
[v1] Thu, 4 Jun 2026 03:50:47 UTC (3,400 KB)
[v2] Fri, 9 Oct 2026 07:41:21 UTC (2,185 KB)