arXiv cs.CY· Naveen Raman, Fei Fang, Yiling Chen·· 11 小时前AI 评分31
LLM 排行榜中的战略竞争与趋同：一项博弈论研究
Strategic Competition and Convergence in LLM Leaderboards
AI 导读
研究将 LLM 排行榜竞争建模为开发者在各项能力上分配资源以最大化胜率的博弈，发现相似的生产函数会导致相似的能力画像（单一化），而比较优势可维持差异化能力画像。基于 LM Arena 和 OpenRouter 数据，开发者在自己落后的类别上进步更快，同时保留类别特定优势，模型前沿日益拥挤且使用模式愈发多样。结果表明竞争加剧未必产生可互换的模型，反而可维持各具优势的竞争性模型。
正文 · 原文
Abstract:Leaderboards are a popular way to compare large language model (LLM) capabilities and can influence how model developers invest resources. We ask when competition on leaderboards leads developers to produce models with similar capability profiles, which we call monoculture, or similar performance levels. We study this by formulating leaderboard competition as a game in which model developers invest resources across capabilities to maximize their win rate. We show that similar production functions can induce similar capability profiles, while comparative advantage can sustain distinct capability profiles. We then study convergence in performance levels and show that diminishing returns can lead to an increasingly crowded competitive frontier. Using data from LM Arena and OpenRouter, we find that developers improve more in categories where they lag competitors, while still retaining category-specific advantages. At the same time, we find that the frontier of models becomes increasingly crowded, alongside increasingly diversified model usage patterns. Our results suggest that increasing competition does not need to produce interchangeable models, but can instead sustain competitive models with distinct strengths.
| Subjects: | Computer Science and Game Theory (cs.GT); Computers and Society (cs.CY); Software Engineering (cs.SE) |
| Cite as: | arXiv:2611.09780 [cs.GT] |
| (or arXiv:2611.09780v1 [cs.GT] for this version) | |
| https://doi.org/10.48550/arXiv.2611.09780 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Naveen Raman [view email]
[v1]
Fri, 6 Nov 2026 17:42:24 UTC (439 KB)
来源：arXiv cs.CY · arxiv.org