Anthropic：模型能力增强后，单代理架构常优于多子代理
原标题：Anthropic says one AI agent beats many subagents now that models can handle complexity alone
AI 摘要
Anthropic 基于多个企业部署经验建议企业评估将多个专用子代理整合为单一主代理，由后者按需加载上下文。其 Applied AI 团队发现，从 Opus 4.5 起的新模型（含 Opus 5.5、Fable 5.1）能在单循环内管理复杂计划，单代理架构在质量、成本和延迟上常优于多代理，某对比中速度约快一倍。Anthropic 在 GitHub 发布迁移工具包，并提醒盲目合并可能破坏原有流程；子代理仍适用于上下文窗口性能明显下降、需并行独立上下文或工具选择困难等场景。
正文节选
Anthropic says one AI agent beats many subagents now that models can handle complexity alone Key Points - Anthropic recommends that companies evaluate whether they can consolidate sub-agents into a single agent that loads context on demand. - The company's experience shows that many specialized sub-agents can lead to higher latency, rising costs, and information loss at every handoff. - Anthropic has published a migration kit on GitHub to help teams make the transition. Anthropic found that a si