Claude Fable 5.2 (Max) 登顶 Agent Arena，10 项类别中 9 项超过 Opus 5.5
Agent Arena 实测榜把 Anthropic 新的 Fable 5.2 (Max) 排到第一，10 项里赢了自己家 Opus 5.5 九项，视觉设计类分数直接翻倍。做前端、写文档的朋友优先试试 Fable，搞金融分析的可以再等等。
Anthropic 的 Claude Fable 5.2 (Max) 在 Agent Arena 排行榜升至第一，在与 Claude Opus 5.5 (Max) 同时上榜的 10 个类别中，9 个类别领先，剩下 1 个为统计意义上的平局。净改进分数上，Fable 5.2 整体为 30.48%，Opus 5.5 为 19.51%。
差距最大的类别是视觉设计：Fable 5.2 达到 45.34%，Opus 5.5 为 23.29%，说明 Fable 在界面审美和视觉产出上明显占优。其他类别中，Documents 为 31.51% 对 17.88%，Knowledge work 为 25.67% 对 13.24%，Code 为 31.45% 对 22.24%。只有 Business & Finance 类别双方分差过小，无法判定胜负。
从数据看，Fable 5.2 的领先并非均匀分布，而是集中在视觉与文档类任务上，代码类优势相对温和。对在 Agent Arena 上选模型做前端、文档或知识工作的开发者来说，Fable 5.2 (Max) 目前是榜单上的首选；而在金融与商业分析场景，两个模型的差距还在误差范围内，可以按价格和延迟自行取舍。
Claude Fable 5.2 (Max) by @AnthropicAI ranks #1 on Agent Arena, ahead of Claude Opus 5.5 (Max) in 9 of the 10 categories where both are ranked. In the 10th category, there is a statistical tie. Where Fable really shines over Opus is its visual taste. Net improvement score for Fable 5.2 vs. Opus 5.5 (both Max) across key categories: - Overall: 30.48% vs. 19.51% - Visual design: 45.34% vs. 23.29% - Documents: 31.51% vs. 17.88% - Knowledge work: 25.67% vs. 13.24% - Code: 31.45% vs. 22.24% In Business & Finance, the gap is too small to call. But the scores only show that Fable is ahead, not where the gap comes from. So we looked at what makes Fable 5.2 different from Opus 5.5 across real-world sessions.📊 22 ⚡
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 US-China foundation model benchmarks and commercialization comparison and AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Claude, Anthropic, Agent Arena, Claude Opus.
Source attribution: curated from lmarena.ai (@lmarena_ai).
