AI RESEARCH

Phase-Scheduled Multi-Agent Systems for Token-Efficient Coordination

arXiv CS.AI

ArXi:2604.17400v1 Announce Type: new Multi-agent systems (MAS) powered by large language models suffer from severe token inefficiency arising from two compounding sources: (i) unstructured parallel execution, where all agents activate simultaneously irrespective of input readiness; and (ii) unrestricted context sharing, where every agent receives the full accumulated context regardless of relevance.