AgentSlimming: Towards Efficient and Cost-Aware Multi-Agent Systems

ACL ARR 2026 January Submission8137 Authors

Keywords: LLM, multi-agent systems, agent pruning, token efficiency

Abstract:

Large Language Model-based Multi-Agent Systems (MAS) have demonstrated remarkable capabilities in complex tasks. However, manually designing optimal communication topologies is labor-intensive, while automated expansion methods often result in bloated structures with redundant agents, leading to excessive token consumption.
To address this problem, we introduce AgentSlimming, a plug-and-play compression framework for graph-structured multi-agent workflows.
Motivated by the AgentPruner and AgentQuant in neural networks, AgentSlimming compresses workflows by firstly estimiate the importance score of each agent with a hybrid mechanism, and then removing redundant agents or replacing them with low-cost ones, where each operation is then validated with a baseline-anchored acceptance rule to prevent performance collapse. Experiments show that AgentSlimming reduces average token cost by up to 78.9% with negligible performance degradation, and even somethimes improves accuracy, achieving a strong Pareto-optimal trade-off between cost and quality.

Paper Type: Long

Research Area: AI/LLM Agents

Research Area Keywords: LLM agents, multi-agent systems, agent communication

Contribution Types: NLP engineering experiment, Approaches low compute settings-efficiency

Languages Studied: English

Submission Number: 8137