md5

作者宣布在 SolveAtHome 平台上线 MD5 挑战项目，邀请任何人贡献闲置 AI tokens，让分布式 AI agent 群体协作探索密码学分析前沿，以检验集体智能能否做出超越单个前沿实验室的科学发现。

This paper tests whether large language models can maintain exact state across long sequences of dependent tool calls by having them compute MD5 hashes step by step. The study demonstrates that with proper context and voting mechanisms, LLMs can achieve high success rates in such exact, multi-step procedures.