Stanford's DeLM Cuts Multi-Agent Costs 50% — No Central Boss Needed
Stanford researchers unveil DeLM, a decentralized language model that reduces multi-agent task costs by 50% by enabling direct agent-to-agent coordination without a central controller.