1000 个 Claude 智能体 11 天用 Lean 形式化费马大定理
Just watched @RLanceMartin (Anthropic) at #AIECode: 1,000 Claude agents formalized Fermat's Last The...
1000 个 Claude 智能体自己拆任务、互相校对，11 天干完了人类团队要干几年的形式化证明，还顺手证明了 3 万多个定理。做智能体编排、Lean 或数学 AI 的朋友都该看看这个案例。
Anthropic 研究员 Lance Martin 在 AIECode 大会披露：1000 个 Claude 智能体协作，用 11 天完成了费马大定理（Fermat's Last Theorem）的 Lean 形式化。
过程中共证明 30,300 个定理，产出 1300 万行 Lean 代码，规模超过 Mathlib 数学库的 5 倍。主智能体在结束时向 main 分支发出冻结指令，并向其余 999 个智能体致谢。
该项目的机制是多智能体分工：主智能体拆解证明依赖，子智能体并行证明各层级引理，再逐步合并。对比此前人类团队（如流体动力学问题形式化项目）耗时数月的进度，这次演示展示了大规模智能体协作在形式化数学上的可行路径。
Just watched @RLanceMartin (Anthropic) at #AIECode: 1,000 Claude agents formalized Fermat's Last The...
Just watched @RLanceMartin (Anthropic) at #AIECode : 1,000 Claude agents formalized Fermat's Last Theorem in 11 days. 30,300 theorems proved along the way. 13M lines of Lean — over 5x the size of Mathlib. The lead agent's last message: "TOTAL FREEZE OF origin/main, EFFECTIVE NOW. … Magnificent work, all 999 of you."📊 1 ⚡
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Claude, Anthropic, Lean, 多智能体.
Source attribution: curated from AI Will (@FinanceYF5). Cross-source corroboration: 11 independent sources have covered this development (including Andrew Curran, 阿里云 Alibaba Cloud, lmarena.ai), confirming industry-wide signal strength.
