Claude Opus · FrontierCode · v1.1 Main set · 历代记录

按发布时间查看 Claude Opus 的 2 次发布在 FrontierCode · v1.1 Main set 的成绩、配置与披露。

仅为直接前代间的唯一、有效精确值计算差值,要求单位、指标与明确记录的配置和条件一致。

Claude Opus 5

· 本次发布有 5 条成绩记录,含 5 条精确值(可含历史记录)

Claude Opus 5 · FrontierCode 成绩与配置
指标与成绩配置与评测条件出处
Overall score53.4%原文 53.4% · 单位 %
公布 未确认
FrontierCode v1.1 Main · Medium effort
设置
Cognition-run FrontierCode evaluation; 150 tasks; overall composite score reported as mean@5. Explicit adaptive-thinking effort: medium.
未知项
The card does not state any additional sampling parameters beyond the cited evaluation description.
工具
工具条件未记录
推理预算
medium
框架
未记录
协议
FrontierCode v1.1 Main; Cognition-scored composite; mean@5.
条件
53.4% is the best Main score at medium effort; use the figure-labeled per-effort rows for the full sweep.
在排名中查看此成绩
Overall score41.9%原文 41.9% · 单位 %
公布 未确认
FrontierCode v1.1 Main · Low effort
设置
Cognition-run FrontierCode evaluation; 150 tasks; overall composite score reported as mean@5. Explicit adaptive-thinking effort: low.
未知项
The card does not state any additional sampling parameters beyond the cited evaluation description.
工具
工具条件未记录
推理预算
low
框架
未记录
协议
FrontierCode v1.1 Main; Cognition-scored overall composite; mean@5.
条件
Each point is an explicit effort-labeled score; the card notes scope-of-task prompting was unchanged from the official evaluation.
在排名中查看此成绩
Overall score48%原文 48% · 单位 %
公布 未确认
FrontierCode v1.1 Main · High effort
设置
Cognition-run FrontierCode evaluation; 150 tasks; overall composite score reported as mean@5. Explicit adaptive-thinking effort: high.
未知项
The card does not state any additional sampling parameters beyond the cited evaluation description.
工具
工具条件未记录
推理预算
high
框架
未记录
协议
FrontierCode v1.1 Main; Cognition-scored overall composite; mean@5.
条件
Each point is an explicit effort-labeled score; the card notes scope-of-task prompting was unchanged from the official evaluation.
在排名中查看此成绩
Overall score43.6%原文 43.6% · 单位 %
公布 未确认
FrontierCode v1.1 Main · Xhigh effort
设置
Cognition-run FrontierCode evaluation; 150 tasks; overall composite score reported as mean@5. Explicit adaptive-thinking effort: xhigh.
未知项
The card does not state any additional sampling parameters beyond the cited evaluation description.
工具
工具条件未记录
推理预算
xhigh
框架
未记录
协议
FrontierCode v1.1 Main; Cognition-scored overall composite; mean@5.
条件
Each point is an explicit effort-labeled score; the card notes scope-of-task prompting was unchanged from the official evaluation.
在排名中查看此成绩
Overall score48%原文 48% · 单位 %
公布 未确认
FrontierCode v1.1 Main · Max effort
设置
Cognition-run FrontierCode evaluation; 150 tasks; overall composite score reported as mean@5. Explicit adaptive-thinking effort: max.
未知项
The card does not state any additional sampling parameters beyond the cited evaluation description.
工具
工具条件未记录
推理预算
max
框架
未记录
协议
FrontierCode v1.1 Main; Cognition-scored overall composite; mean@5.
条件
Each point is an explicit effort-labeled score; the card notes scope-of-task prompting was unchanged from the official evaluation.
在排名中查看此成绩

Claude Opus 5.5

· 本次发布有 2 条成绩记录,含 2 条精确值(可含历史记录)

与直接前代的记录缺少满足差值计算的同条件依据。

Claude Opus 5.5 · FrontierCode 成绩与配置
指标与成绩配置与评测条件出处
Overall score54.4%原文 54.4% · 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
Claude Code
协议
FrontierCode v1.1 Main。
条件
除特别注明外 adaptive thinking/max effort。 Source-wide conditions: Cognition ran the evaluations in Claude Code; mean@5. Main and Extended are separate benchmark definitions.
在排名中查看此成绩
Overall score54.6%原文 54.6% · 单位 %
公布 未确认
FrontierCode v1.1 Main medium effort
设置
Explicit medium effort; Cognition evaluation in Claude Code. Mean@5.
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
Cognition Claude Code harness
协议
FrontierCode v1.1 Main; Cognition Claude Code harness; mean@5; medium effort.
条件
Best reported effort. Evaluation has internet access; no timeout information is stated. Do not infer effort from output tokens. Source-wide conditions: Cognition ran the evaluations in Claude Code; mean@5. Main and Extended are separate benchmark definitions.
在排名中查看此成绩