Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified) · 附件证据(ev-f52-automationbench-card)

Claude Fable 5.2 的 AutomationBench · 1.0.6; 47 tools; chart-reported score, scoring rule unspecified 官方出处。见《Claude Fable 5.2 and Claude Mythos 5.2 System Card》PDF p.240, Figure 8.14.5.A, AutomationBench 1.0.6(已渲染并视觉核读数值标签)。证据 ev-f52-automationbench-card。

原文与定位

Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Adaptive Max配置,Score 51.4%(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.240, Figure 8.14.5.A, AutomationBench 1.0.6(已渲染并视觉核读数值标签)。 查看成绩条件

Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Low effort配置,Score 39.7%(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.240, Figure 8.14.5.A, AutomationBench 1.0.6(已渲染并视觉核读数值标签)。 查看成绩条件

Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Medium effort配置,Score 40.2%(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.240, Figure 8.14.5.A, AutomationBench 1.0.6(已渲染并视觉核读数值标签)。 查看成绩条件

Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),High effort配置,Score 44.3%(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.240, Figure 8.14.5.A, AutomationBench 1.0.6(已渲染并视觉核读数值标签)。 查看成绩条件

Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Extra high effort配置,Score 47.9%(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.240, Figure 8.14.5.A, AutomationBench 1.0.6(已渲染并视觉核读数值标签)。 查看成绩条件

Zapier private set, 657 tasks per effort, API default fallbacks enabled; Fable 5.2 scores are labeled 39.7% low, 40.2% medium, 44.3% high, 47.9% extra high, and 51.4% max. The Fable run was October 8, 2026.
官方材料
Claude Fable 5.2 and Claude Mythos 5.2 System Card · PDF 第 240 页
定位
附件 · PDF p.240, Figure 8.14.5.A, AutomationBench 1.0.6(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 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Adaptive Max配置,Score 51.4%(具体成绩公布时间未确认);Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Low effort配置,Score 39.7%(具体成绩公布时间未确认);Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Medium effort配置,Score 40.2%(具体成绩公布时间未确认);Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),High effort配置,Score 44.3%(具体成绩公布时间未确认);Claude Fable 5.2 · AutomationBench(1.0.6; 47 tools; chart-reported score, scoring rule unspecified),Extra high effort配置,Score 47.9%(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.240, Figure 8.14.5.A, AutomationBench 1.0.6(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=240。Benchlog https://benchlog.lekshome.cc/evidence/claude-fable-and-mythos-5-2/ev-f52-automationbench-card/

关联成绩与条件

Claude Fable 5.2 · AutomationBench 成绩与配置
指标与成绩配置与评测条件证据关系
Score51.4%原文 51.4% · 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
工具条件未记录
推理预算
max
框架
未记录
协议
AutomationBench 1.0.6 private held-out set; 657 tasks per effort level; Zapier API default fallbacks; max effort.
条件
Fable 5.2 score/cost based on Zapier run of October 8, 2026; cost is not included in this result.

直接支持该成绩

在排名中查看此成绩
Score39.7%原文 39.7% · 单位 %
公布 未确认
Low effort
设置
Claude Fable 5.2;adaptive thinking always on;Low effort。
未知项
未另记
工具
工具条件未记录
推理预算
low
框架
未记录
协议
AutomationBench 1.0.6 private held-out set; 657 tasks per effort level; low effort; API default fallbacks.
条件
Zapier run dated October 8, 2026.

直接支持该成绩

在排名中查看此成绩
Score40.2%原文 40.2% · 单位 %
公布 未确认
Medium effort
设置
Claude Fable 5.2;adaptive thinking always on;Medium effort。
未知项
主文未列该配置的采样设置与运行数。
工具
工具条件未记录
推理预算
medium
框架
未记录
协议
AutomationBench 1.0.6 private held-out set; 657 tasks per effort level; medium effort; API default fallbacks.
条件
Zapier run dated October 8, 2026.

直接支持该成绩

在排名中查看此成绩
Score44.3%原文 44.3% · 单位 %
公布 未确认
High effort
设置
Claude Fable 5.2;adaptive thinking always on;High effort。
未知项
主文未列该配置的采样设置与运行数。
工具
工具条件未记录
推理预算
high
框架
未记录
协议
AutomationBench 1.0.6 private held-out set; 657 tasks per effort level; high effort; API default fallbacks.
条件
Zapier run dated October 8, 2026.

直接支持该成绩

在排名中查看此成绩
Score47.9%原文 47.9% · 单位 %
公布 未确认
Extra high effort
设置
Claude Fable 5.2;AutomationBench 1.0.6 的 Extra high effort。
未知项
该 effort 标签按 Zapier 原图记录;不与 Xhigh 合并。
工具
工具条件未记录
推理预算
extra-high
框架
未记录
协议
AutomationBench 1.0.6 private held-out set; 657 tasks per effort level; extra-high effort; API default fallbacks.
条件
Zapier run dated October 8, 2026.

直接支持该成绩