Saved in:
Bibliographic Details
Main Authors: Tian, Aaron Xuxiang, Zhang, Ruofan, Tang, Jiayao, Cho, Young Min, Li, Xueqian, Yi, Qiang, Wang, Ji, Zhang, Zhunping, Qi, Danrui, Li, Zekun, Xiang, Xingyu, Guntuku, Sharath Chandra, Ungar, Lyle, Shi, Tianyu, Wang, Chi
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.23537
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912622166671360
author Tian, Aaron Xuxiang
Zhang, Ruofan
Tang, Jiayao
Cho, Young Min
Li, Xueqian
Yi, Qiang
Wang, Ji
Zhang, Zhunping
Qi, Danrui
Li, Zekun
Xiang, Xingyu
Guntuku, Sharath Chandra
Ungar, Lyle
Shi, Tianyu
Wang, Chi
author_facet Tian, Aaron Xuxiang
Zhang, Ruofan
Tang, Jiayao
Cho, Young Min
Li, Xueqian
Yi, Qiang
Wang, Ji
Zhang, Zhunping
Qi, Danrui
Li, Zekun
Xiang, Xingyu
Guntuku, Sharath Chandra
Ungar, Lyle
Shi, Tianyu
Wang, Chi
contents We study multi-turn multi-agent orchestration, where multiple large language model (LLM) agents interact over multiple turns by iteratively proposing answers or casting votes until reaching consensus. Using four LLMs (Gemini 2.5 Pro, GPT-5, Grok 4, and Claude Sonnet 4) on GPQA-Diamond, IFEval, and MuSR, we conduct two experiments: (i) benchmarking orchestration against single-LLM baselines; and (ii) ablations on GPQA-Diamond that vary whether agents see who authored answers and whether they can observe ongoing votes. Orchestration matches or exceeds the strongest single model and consistently outperforms the others. Analysis of best-achievable orchestration performance shows potential for further gains. The ablations show that revealing authorship increases self-voting and ties, and that showing ongoing votes amplifies herding, which speeds convergence but can sometimes yield premature consensus.
format Preprint
id arxiv_https___arxiv_org_abs_2509_23537
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond the Strongest LLM: Multi-Turn Multi-Agent Orchestration vs. Single LLMs on Benchmarks
Tian, Aaron Xuxiang
Zhang, Ruofan
Tang, Jiayao
Cho, Young Min
Li, Xueqian
Yi, Qiang
Wang, Ji
Zhang, Zhunping
Qi, Danrui
Li, Zekun
Xiang, Xingyu
Guntuku, Sharath Chandra
Ungar, Lyle
Shi, Tianyu
Wang, Chi
Artificial Intelligence
We study multi-turn multi-agent orchestration, where multiple large language model (LLM) agents interact over multiple turns by iteratively proposing answers or casting votes until reaching consensus. Using four LLMs (Gemini 2.5 Pro, GPT-5, Grok 4, and Claude Sonnet 4) on GPQA-Diamond, IFEval, and MuSR, we conduct two experiments: (i) benchmarking orchestration against single-LLM baselines; and (ii) ablations on GPQA-Diamond that vary whether agents see who authored answers and whether they can observe ongoing votes. Orchestration matches or exceeds the strongest single model and consistently outperforms the others. Analysis of best-achievable orchestration performance shows potential for further gains. The ablations show that revealing authorship increases self-voting and ties, and that showing ongoing votes amplifies herding, which speeds convergence but can sometimes yield premature consensus.
title Beyond the Strongest LLM: Multi-Turn Multi-Agent Orchestration vs. Single LLMs on Benchmarks
topic Artificial Intelligence
url https://arxiv.org/abs/2509.23537