Rethinking the Value of Multi-Agent Workflow: A Strong Single Agent Baseline
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915737641156608 |
|---|---|
| author | Xu, Jiawei Koesdwiady, Arief Bei, Sisong Han, Yan Huang, Baixiang Wang, Dakuo Chen, Yutong Wang, Zheshen Wang, Peihao Li, Pan Ding, Ying |
| author_facet | Xu, Jiawei Koesdwiady, Arief Bei, Sisong Han, Yan Huang, Baixiang Wang, Dakuo Chen, Yutong Wang, Zheshen Wang, Peihao Li, Pan Ding, Ying |
| contents | Recent advances in LLM-based multi-agent systems (MAS) show that workflows composed of multiple LLM agents with distinct roles, tools, and communication patterns can outperform single-LLM baselines on complex tasks. However, most frameworks are homogeneous, where all agents share the same base LLM and differ only in prompts, tools, and positions in the workflow. This raises the question of whether such workflows can be simulated by a single agent through multi-turn conversations. We investigate this across seven benchmarks spanning coding, mathematics, general question answering, domain-specific reasoning, and real-world planning and tool use. Our results show that a single agent can reach the performance of homogeneous workflows with an efficiency advantage from KV cache reuse, and can even match the performance of an automatically optimized heterogeneous workflow. Building on this finding, we propose \textbf{OneFlow}, an algorithm that automatically tailors workflows for single-agent execution, reducing inference costs compared to existing automatic multi-agent design frameworks without trading off accuracy. These results position the single-LLM implementation of multi-agent workflows as a strong baseline for MAS research. We also note that single-LLM methods cannot capture heterogeneous workflows due to the lack of KV cache sharing across different LLMs, highlighting future opportunities in developing \textit{truly} heterogeneous multi-agent systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_12307 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Rethinking the Value of Multi-Agent Workflow: A Strong Single Agent Baseline Xu, Jiawei Koesdwiady, Arief Bei, Sisong Han, Yan Huang, Baixiang Wang, Dakuo Chen, Yutong Wang, Zheshen Wang, Peihao Li, Pan Ding, Ying Multiagent Systems Computation and Language Machine Learning Recent advances in LLM-based multi-agent systems (MAS) show that workflows composed of multiple LLM agents with distinct roles, tools, and communication patterns can outperform single-LLM baselines on complex tasks. However, most frameworks are homogeneous, where all agents share the same base LLM and differ only in prompts, tools, and positions in the workflow. This raises the question of whether such workflows can be simulated by a single agent through multi-turn conversations. We investigate this across seven benchmarks spanning coding, mathematics, general question answering, domain-specific reasoning, and real-world planning and tool use. Our results show that a single agent can reach the performance of homogeneous workflows with an efficiency advantage from KV cache reuse, and can even match the performance of an automatically optimized heterogeneous workflow. Building on this finding, we propose \textbf{OneFlow}, an algorithm that automatically tailors workflows for single-agent execution, reducing inference costs compared to existing automatic multi-agent design frameworks without trading off accuracy. These results position the single-LLM implementation of multi-agent workflows as a strong baseline for MAS research. We also note that single-LLM methods cannot capture heterogeneous workflows due to the lack of KV cache sharing across different LLMs, highlighting future opportunities in developing \textit{truly} heterogeneous multi-agent systems. |
| title | Rethinking the Value of Multi-Agent Workflow: A Strong Single Agent Baseline |
| topic | Multiagent Systems Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2601.12307 |