arxivcs.MAcs.AIcs.CL2026-07-06
Decision Protocols in Multi-Agent Large Language Model Conversations
Improving the task performance of Large Language Models (LLMs) is essential, yet scaling these models faces significant challenges such as diminishing returns and high costs. Multi-Agent Systems (MAS) offer a promising solution by distributing tasks among specialized agents to im…