MANTA: Multi-Agent Network Topology Adaptation for Self-Evolving Multi-Agent Systems

2026-07-30

AI Summary

This work addresses the limitation of existing large language model–based multi-agent systems, which typically rely on static communication topologies and struggle to adapt dynamically to task demands during inference. To overcome this, the authors propose MANTA, a framework that enables bounded adaptive adjustments—during inference—to role assignment, communication links, execution order, information visibility, and verification paths, all while preserving the original task interface and agent budget. MANTA integrates task-conditioned topology initialization, collaborative trajectory monitoring, and structural update mechanisms to support dynamic evolution of multi-agent organizational structures. Evaluated across five benchmark tasks, MANTA achieves an average score of 74.0, outperforming the strongest baseline by 5.8 percentage points and attaining state-of-the-art performance on the PlanCraft task.

Abstract

Large language model-based multi-agent systems improve complex problem solving through task decomposition, agent specialization, information exchange, and intermediate validation. However, existing systems typically treat communication topology as a fixed design choice or an offline optimization target. We introduce MANTA, a framework for Multi-Agent Network Topology Adaptation that enables communication structures to self-evolve at inference time. Before execution, MANTA initializes a task-conditioned topology from prior structural experience. During deployment, it monitors collaboration traces and applies bounded structural updates when the current organization becomes insufficient. These updates can modify agent roles, communication links, execution order, information visibility, and validation pathways while preserving the task interface and agent budget. We evaluate MANTA against representative single-agent and multi-agent baselines on five benchmarks spanning information seeking, tool use, planning, workflow execution, and mathematical reasoning. MANTA achieves the highest average score of 74.0, outperforming the strongest baseline by 5.8 percentage points and obtaining the best result on PlanCraft. These results show that inference-time self-improvement can extend to the architecture of collaboration itself.

Problem

• multi-agent systems
• communication topology
• self-evolution
• inference-time adaptation
• collaboration architecture

Innovation

• Multi-Agent Systems
• Network Topology Adaptation
• Self-Evolving Architecture
• Inference-Time Adaptation
• Collaborative Reasoning