研究：多智能体框架多数跑不赢单智能体加 CoT-SC
Salesforce Research 实测 6 个多智能体框架，发现多数不如单智能体加 CoT-SC，成本还贵 10 倍，做 Agent 前先看看这篇。
Salesforce Research 测试了 6 个自动多智能体框架、4 个 LLM 和 5 个基准，发现它们整体不如 Chain-of-Thought 加 Self-Consistency（CoT-SC），成本却高出最多 10 倍。单智能体 GPT-5 用 CoT-SC 能以不到一半的 token 量击败基于 GPT-4o 的最复杂多智能体框架。但在团队自建的合成多跳金融推理基准 SMFR 上，专家设计的多智能体系统把 GPT-5 从 57.0% 提到 96.5%，成本与 CoT-SC 相当。结论是：多智能体只在子任务可并行、上下文可分离时才有优势。
The illusion of multi-agent advantage: https://t.co/7BkU0h0KYG
Do multi-agent systems outperform a single agent? Across six automatic multi-agent frameworks, four LLMs, and five benchmarks, we found they consistently underperform Chain-of-Thought with Self-Consistency (CoT-SC) while costing up to 10x more.
→ CoT-SC often reaches higher accuracy at under 10% of the compute cost → A single-agent GPT-5 with CoT-SC beats the most sophisticated GPT-4o-based multi-agent frameworks using less than half the tokens → An expert-designed multi-agent system lifts GPT-5 from 57.0% to 96.5% on our new synthetic multi-hop financial reasoning (SMFR) benchmark, at cost comparable to CoT-SC
Multi-agent systems deliver when they are deliberately structured around parallelizable sub-tasks and context separation.
#AgenticAI #EnterpriseAI
