Do next-generation LLM agents inherit the cooperative biases documented in their predecessors, or does scale and provider diversity reshape equilibrium behaviour in competitive multi-agent settings? Willis et al. established a benchmark for this question using evolutionary game theory and the Iterated Prisoner's Dilemma (IPD), finding consistent cooperative biases in ChatGPT-4o and Claude 3.5 Sonnet. We extend this benchmark to four frontier models released in 2025-2026 - Claude Sonnet 4.6, Gemini 2.5 Flash, Gemini 3.1 Pro, and GPT-5.4 Mini - applying the identical protocol across three prompting styles (Default, Prose, Self-Refine) and four population compositions (balanced and biased, with and without noise). Cooperative bias persists across providers (H1): ten of twelve model-prompt combinations favour cooperative equilibria in balanced noiseless conditions. Cross-provider divergence is substantial (H3): Gemini 2.5 Flash reaches up to 77% aggressive equilibria under biased conditions, while GPT-5.4 Mini reaches 70% cooperative equilibria under Self-Refine. Support for aggressive capability parity is partial (H2): Self-Refine raises ICD in all models and Gemini 3.1 Pro Refine achieves the highest ICD in the dataset (0.925), but Default and Prose prompts show no systematic narrowing. Evidence on noise robustness is directionally positive but not robustly confirmed (H4): with n=500 Moran iterations per condition, average noise sensitivity is about 6 percentage points for Claude Sonnet 4.6 versus 13 pp for Claude 3.5 Sonnet, but this cross-study gap is not statistically significant once the predecessor's unreported sampling error is propagated. Provider identity, rather than model generation, is the strongest correlate of equilibrium outcomes; noise remains a universal challenge regardless of model size or vintage.

翻译：下一代LLM智能体是否会继承其前身所记录的合作偏差？还是规模和提供商的多样性会重塑竞争性多智能体设定中的均衡行为？Willis等人使用进化博弈论和迭代囚徒困境（IPD）为这一问题建立了基准，发现ChatGPT-4o和Claude 3.5 Sonnet中存在一致的合作偏差。我们将此基准扩展到2025-2026年发布的四个前沿模型——Claude Sonnet 4.6、Gemini 2.5 Flash、Gemini 3.1 Pro和GPT-5.4 Mini——在三种提示风格（默认、散文、自我精炼）和四种群体组成（平衡与有偏，有无噪声）下应用相同的协议。合作偏差在提供商间持续存在（H1）：十二种模型-提示组合中的十种在平衡无噪声条件下偏向合作均衡。跨提供商差异显著（H3）：在有偏条件下，Gemini 2.5 Flash达到高达77%的侵略性均衡，而GPT-5.4 Mini在自我精炼下达到70%的合作均衡。对侵略性能力对等性的支持是部分性的（H2）：自我精炼提高了所有模型的ICD，且Gemini 3.1 Pro精炼在数据集中实现了最高的ICD（0.925），但默认和散文提示未显示系统性缩小。关于噪声鲁棒性的证据方向积极但未得到稳健确认（H4）：在每个条件下进行n=500次Moran迭代后，Claude Sonnet 4.6的平均噪声敏感度约为6个百分点，而Claude 3.5 Sonnet为13个百分点，但一旦传播其前身未报告的抽样误差，这一跨研究差距在统计上不显著。提供商身份，而非模型代际，是均衡结果的最强相关性因素；无论模型大小或年代，噪声仍然是一个普遍挑战。