Claude Opus 4.5 · SpreadsheetBench(Full 912-problem set) 64.25% · 附件证据(ev-op45-card-spreadsheet)

Claude Opus 4.5 的 SpreadsheetBench · Full 912-problem set 官方出处。SpreadsheetBench, no extended thinking · 64.25%。见《Claude Opus 4.5 System Card》PDF p.31, §2.15 SpreadsheetBench。证据 ev-op45-card-spreadsheet。

原文与定位

Claude Opus 4.5 · SpreadsheetBench(Full 912-problem set),SpreadsheetBench, no extended thinking配置,Score 64.25%(具体成绩公布时间未确认)。见《Claude Opus 4.5 System Card》PDF p.31, §2.15 SpreadsheetBench。 查看成绩条件

We used the full 912-problem set. ... With no extended thinking, and a 200k context window, Claude Opus 4.5 achieved a score of 64.25% on SpreadsheetBench (averaged across 5 trials).
官方材料
Claude Opus 4.5 System Card · PDF 第 31 页
定位
附件 · PDF p.31, §2.15 SpreadsheetBench
材料版本
Current saved PDF/TXT: 本地留存副本 and .txt; cover November 2025, change log 2025-11-24; current binary is not established as launch-day bytes; fully reviewed 2026-11-16.
版本公布
未确认
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 Opus 4.5 · SpreadsheetBench(Full 912-problem set),SpreadsheetBench, no extended thinking配置,Score 64.25%(具体成绩公布时间未确认)。官方出处 Claude Opus 4.5 System Card(附件,Current saved PDF/TXT: 本地留存副本 and .txt; cover November 2025, change log 2025-11-24; current binary is not established as launch-day bytes; fully reviewed 2026-11-16.)PDF p.31, §2.15 SpreadsheetBench,https://www.anthropic.com/claude-opus-4-5-system-card#page=31。Benchlog https://benchlog.lekshome.cc/evidence/claude-opus-4-5/ev-op45-card-spreadsheet/

关联成绩与条件

Claude Opus 4.5 · SpreadsheetBench 成绩与配置
指标与成绩配置与评测条件证据关系
Score64.25%原文 64.25% · 单位 %
公布 未确认
SpreadsheetBench, no extended thinking
设置
Full 912-problem set; no extended thinking; 200K context; custom bash/string/Python harness; five-trial average.
未知项
No further prompt or sampling details are stated.
工具
允许工具
推理预算
未记录
框架
未记录
协议
SpreadsheetBench full 912-problem set; custom bash/string editing/Python harness; five-trial average.
条件
No extended thinking; 200K context. Python environment included openpyxl, LibreOffice, pandas and NumPy.

直接支持该成绩

在排名中查看此成绩