Optimizing Sequential Multi-Step Tasks with Parallel LLM Agents

2025-07-11

AI Summary

To address high latency and low task completion rates in LLM-based multi-agent systems caused by serial reasoning in complex, multi-step tasks, this paper proposes M1-Parallel—a framework enabling parallel exploration of diverse solution paths via event-driven asynchronous communication among multiple LLM agent teams, followed by result fusion for enhanced robustness. Its key innovation lies in modeling plan diversity as a parallel search space, integrated with lightweight scheduling and asynchronous message passing to achieve end-to-end acceleration without compromising accuracy. Experimental evaluation on representative complex tasks demonstrates that M1-Parallel achieves up to 2.2× speedup and a 37.5% improvement in task completion rate over state-of-the-art baselines, significantly outperforming both sequential and coarse-grained parallel approaches.

Technology Category

• Multiagent Systems: Multiagent Planning
• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• 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
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Large language model (LLM)-based multi-agent systems have demonstrated remarkable promise for tackling complex tasks by breaking them down into subtasks that are iteratively planned, executed, observed, and refined. Despite their effectiveness, these systems often incur high latency because real-world problems frequently demand multiple iterative cycles of reasoning steps. To address this challenge, we propose M1-Parallel, a framework that concurrently runs multiple multi-agent teams in parallel to uncover distinct solution paths. By leveraging an event-driven communication model with asynchronous messaging, M1-Parallel efficiently capitalizes on the inherent diversity of valid plans to either reduce end-to-end latency or boost task completion rates. Our experiments on complex tasks show that M1-Parallel with early termination achieves up to $2.2 imes$ speedup while preserving accuracy, and that M1-Parallel with aggregation yields higher task completion rates. We further investigate strategies aimed at encouraging diverse execution plans but observe no additional performance gains over repeated sampling. Overall, these findings underscore the potential of parallel plan execution for optimizing multi-agent systems for real-world, high-complexity reasoning tasks.

Problem

• Reducing latency in multi-agent LLM systems
• Enhancing task completion rates via parallel execution
• Optimizing complex reasoning with diverse solution paths

Innovation

• Parallel multi-agent teams for diverse solutions
• Event-driven async messaging for efficiency
• Early termination and aggregation strategies