Claude Fable 5.2 · Multi-task Indic Language Understanding Benchmark(10 Indic languages plus English · overall average) 94.1% · 附件证据(ev-f52-milu)

Claude Fable 5.2 的 Multi-task Indic Language Understanding Benchmark · 10 Indic languages plus English · overall average 官方出处。Adaptive Max · 94.1%。见《Claude Fable 5.2 and Claude Mythos 5.2 System Card》PDF p.249, Figure 8.17.2.A, MILU(已渲染并视觉核读数值标签)。证据 ev-f52-milu。

原文与定位

Claude Fable 5.2 · Multi-task Indic Language Understanding Benchmark(10 Indic languages plus English · overall average),Adaptive Max配置,Average accuracy 94.1%(具体成绩公布时间未确认)。见《Claude Fable 5.2 and Claude Mythos 5.2 System Card》PDF p.249, Figure 8.17.2.A, MILU(已渲染并视觉核读数值标签)。 查看成绩条件

Fable 5.2 achieved 94.1% average accuracy across 11 languages; max effort, safety classifiers enabled, no tools, five-trial average; blocked/unanswered items excluded.
官方材料
Claude Fable 5.2 and Claude Mythos 5.2 System Card · PDF 第 249 页
定位
附件 · PDF p.249, Figure 8.17.2.A, MILU(已渲染并视觉核读数值标签)
材料版本
Official System Card; 264-page PDF, cover dated November 9, 2026; current PDF/TXT saved as 本地留存副本 and .txt; PDF pp.1-264 text read page-by-page; all 80 image/vector-figure pages in pp.1-206 were visually checked in 21 sheets, and capability figures in pp.207-254 were visually checked; no unread pages or figures remain.
版本公布
披露时间
未确认 · 首发/后补未确认
采集时间
历史依据
暂无可靠历史版本依据
引用文本

Claude Fable 5.2 · Multi-task Indic Language Understanding Benchmark(10 Indic languages plus English · overall average),Adaptive Max配置,Average accuracy 94.1%(具体成绩公布时间未确认)。官方出处 Claude Fable 5.2 and Claude Mythos 5.2 System Card(附件,Official System Card; 264-page PDF, cover dated November 9, 2026; current PDF/TXT saved as 本地留存副本 and .txt; PDF pp.1-264 text read page-by-page; all 80 image/vector-figure pages in pp.1-206 were visually checked in 21 sheets, and capability figures in pp.207-254 were visually checked; no unread pages or figures remain.)PDF p.249, Figure 8.17.2.A, MILU(已渲染并视觉核读数值标签),https://www.anthropic.com/document/claude-fable-5-2-mythos-5-2-system-card#page=249。Benchlog https://benchlog.lekshome.cc/evidence/claude-fable-and-mythos-5-2/ev-f52-milu/

关联成绩与条件

Claude Fable 5.2 · Multi-task Indic Language Understanding Benchmark 成绩与配置
指标与成绩配置与评测条件证据关系
Average accuracy94.1%原文 94.1% · 单位 %
公布 未确认
Adaptive Max
设置
Claude Fable 5.2;adaptive thinking always on;Max effort;默认 temperature/top_p;Card 通用表说明通常五次运行平均,个别测试另行注明。
未知项
System Card pp.1-264 文字已完整核读;各benchmark章节未明确披露的运行参数仍未知。
工具
禁止工具
推理预算
max
框架
未记录
协议
MILU average accuracy across 11 languages; max effort; five-trial average; no tools or custom system prompt.
条件
Safety classifiers enabled; blocked/unanswered questions under 0.2% excluded from average.

直接支持该成绩

在排名中查看此成绩