中国前沿大模型合作行为并非铁板一块：实验室间差异显著
原标题：Not a Monolith: Lab-Level Divergence in the Cooperative Equilibria of Chinese Frontier LLM Agents
AI 摘要
本研究通过进化迭代囚徒困境实验，比较了四款中国前沿大模型（DeepSeek V4 Pro、Qwen3-Max、Kimi K2.5、GLM-5.1）的合作倾向。研究发现，这些模型并非铁板一块，实验室间差异显著，且中国模型内部的差异大于中国与西方生态系统之间的平均差异。研究还发现，中国模型的合作倾向普遍性较弱，但这一结论受转换器选择影响。
正文节选
Not a Monolith: Lab-Level Divergence in the Cooperative Equilibria of Chinese Frontier LLM Agents Abstract. Does the cooperative bias documented for Western frontier llm agents extend to a different alignment lineage, and should the Chinese models that embody it be treated as a single bloc or as distinct laboratories? We answer both questions with an evolutionary Iterated Prisoner’s Dilemma study of four frontier-tier Chinese models—DeepSeek V4 Pro, Qwen3-Max, Kimi K2.5, and GLM-5.1—under a desi