Rethinking the Value of Multi-Agent Workflow: A Strong Single Agent Baseline

2026-01-18

AI Summary

This work investigates whether homogeneous multi-agent workflows can be efficiently simulated by a single agent through multi-turn dialogue to reduce system complexity. To this end, the authors propose OneFlow, an algorithm that automatically transforms multi-agent workflows into a single-agent execution scheme leveraging KV cache reuse. The approach is systematically evaluated across seven task categories, including coding, mathematical reasoning, and question answering. Experimental results demonstrate that the single-agent formulation achieves comparable accuracy while significantly outperforming both existing homogeneous and automatically optimized heterogeneous multi-agent systems in inference efficiency. These findings validate the proposed method as a strong baseline and reveal that homogeneous multi-agent systems can be effectively and efficiently reduced to a single-agent implementation.

Technology Category

• Multiagent Systems: Teamwork
• Search and Optimization: Sampling/Simulation-based Search
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search

Abstract

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.

Problem

• multi-agent systems
• single-agent baseline
• homogeneous workflows
• LLM-based agents
• workflow simulation

Innovation

• multi-agent systems
• single-agent baseline
• KV cache reuse
• OneFlow
• homogeneous workflows