Claude Opus 5.5 · CursorBench(4.0) · 附件证据(ev-op55-card-cursorbench)

Claude Opus 5.5 的 CursorBench · 4.0 官方出处。见《Claude Opus 5.5 System Card》§8.8, pp.179-180, Figure 8.8.A。证据 ev-op55-card-cursorbench。

原文与定位

Claude Opus 5.5 · CursorBench(4.0),Adaptive Max配置,Score 57.8%(具体成绩公布时间未确认)。见《Claude Opus 5.5 System Card》§8.8, pp.179-180, Figure 8.8.A。 查看成绩条件

Claude Opus 5.5 · CursorBench(4.0),CursorBench 4.0 medium effort配置,Score 52.5%(具体成绩公布时间未确认)。见《Claude Opus 5.5 System Card》§8.8, pp.179-180, Figure 8.8.A。 查看成绩条件

Claude Opus 5.5 · CursorBench(4.0),CursorBench 4.0 high effort配置,Score 56%(具体成绩公布时间未确认)。见《Claude Opus 5.5 System Card》§8.8, pp.179-180, Figure 8.8.A。 查看成绩条件

Claude Opus 5.5 · CursorBench(4.0),CursorBench 4.0 xhigh effort配置,Score 56%(具体成绩公布时间未确认)。见《Claude Opus 5.5 System Card》§8.8, pp.179-180, Figure 8.8.A。 查看成绩条件

Opus 5.5 scored 57.8% on CursorBench at max effort, 56.0% at xhigh and high, and 52.5% at medium.
官方材料
Claude Opus 5.5 System Card · PDF 第 179 页
定位
附件 · §8.8, pp.179-180, Figure 8.8.A
材料版本
官方当前 System Card;封面日期 2026-09-22;本地 PDF/TXT:本地留存副本、current-opus-5-5-system-card.txt;230 页全文核读;当前修订时间未知
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 5.5 · CursorBench(4.0),Adaptive Max配置,Score 57.8%(具体成绩公布时间未确认);Claude Opus 5.5 · CursorBench(4.0),CursorBench 4.0 medium effort配置,Score 52.5%(具体成绩公布时间未确认);Claude Opus 5.5 · CursorBench(4.0),CursorBench 4.0 high effort配置,Score 56%(具体成绩公布时间未确认);Claude Opus 5.5 · CursorBench(4.0),CursorBench 4.0 xhigh effort配置,Score 56%(具体成绩公布时间未确认)。官方出处 Claude Opus 5.5 System Card(附件,官方当前 System Card;封面日期 2026-09-22;本地 PDF/TXT:本地留存副本、current-opus-5-5-system-card.txt;230 页全文核读;当前修订时间未知)§8.8, pp.179-180, Figure 8.8.A,https://www.anthropic.com/claude-opus-5-5-system-card#page=179。Benchlog https://benchlog.lekshome.cc/evidence/claude-opus-5-5/ev-op55-card-cursorbench/

关联成绩与条件

Claude Opus 5.5 · CursorBench 成绩与配置
指标与成绩配置与评测条件证据关系
Score57.8%原文 57.8% · 单位 %
公布 未确认
Adaptive Max
设置
官方主表标准配置为 adaptive thinking/max effort、默认 temperature/top_p;具体工具、safeguards、fallback 与运行数按各评测章节记录,不将其一概视为生产 safeguards enabled。
未知项
表格没有为每个 benchmark 单独列出采样设置和运行数;章节明确列出的差异优先。
工具
工具条件未记录
推理预算
max
框架
Cursor production agent harness
协议
CursorBench 4.0。
条件
除特别注明外 adaptive thinking/max effort。 Source-wide conditions: Cursor production agent harness; Cursor independently measured scores; costs are estimated from reported token counts and list prices.

直接支持该成绩

在排名中查看此成绩
Score52.5%原文 52.5% · 单位 %
公布 未确认
CursorBench 4.0 medium effort
设置
Explicit medium effort in Cursor's production agent harness; Cursor measured the score.
未知项
No API-equivalent harness or run date is stated.
工具
工具条件未记录
推理预算
medium
框架
Cursor production agent harness
协议
CursorBench 4.0; Cursor production agent harness; medium effort.
条件
Cursor's published score; no uncertainty interval or run date stated. Source-wide conditions: Cursor production agent harness; Cursor independently measured scores; costs are estimated from reported token counts and list prices.

直接支持该成绩

在排名中查看此成绩
Score56%原文 56.0% · 单位 %
公布 未确认
CursorBench 4.0 high effort
设置
Explicit high effort in Cursor's production agent harness; Cursor measured the score.
未知项
No API-equivalent harness or run date is stated.
工具
工具条件未记录
推理预算
high
框架
Cursor production agent harness
协议
CursorBench 4.0; Cursor production agent harness; high effort.
条件
Cursor's published score; same value as xhigh at a separate effort setting. Source-wide conditions: Cursor production agent harness; Cursor independently measured scores; costs are estimated from reported token counts and list prices.

直接支持该成绩

在排名中查看此成绩
Score56%原文 56.0% · 单位 %
公布 未确认
CursorBench 4.0 xhigh effort
设置
Explicit xhigh effort in Cursor's production agent harness; Cursor measured the score.
未知项
No API-equivalent harness or run date is stated.
工具
工具条件未记录
推理预算
xhigh
框架
Cursor production agent harness
协议
CursorBench 4.0; Cursor production agent harness; xhigh effort.
条件
Cursor's published score; same value as high at a separate effort setting. Source-wide conditions: Cursor production agent harness; Cursor independently measured scores; costs are estimated from reported token counts and list prices.

直接支持该成绩

在排名中查看此成绩