GPT-5.5 · GeneBench · 2026-04-23 report; 103 evaluations across 10 domains · 成绩与配置

GPT-5.5 的 GeneBench · 2026-04-23 report; 103 evaluations across 10 domains 记录。本次发布有 7 条成绩记录,含 7 条精确值(可含历史记录),附 15 条来源定位。

成绩与评测条件

GPT-5.5 · GeneBench 成绩与配置
指标与成绩配置与评测条件出处
Mean pass rate25%原文 25.0% · 单位 %
公布 未确认
GPT-5.5 xhigh
设置
Reasoning effort xhigh; research environment; output may differ slightly from production ChatGPT in some cases.
未知项
A subset of protocol details remain benchmark-specific and unreported.
工具
允许工具
推理预算
xhigh
框架
未记录
协议
GeneBench full suite; 103 evaluations across 10 domains; mean unweighted per-problem pass rate over repeated runs; 95% CI in source excerpt
条件
Mainline GPT-5.5 evaluation at xhigh in research environment.; Reasoning effort xhigh; report states GPT-5.5 xhigh for the mainline run.; GPT-5.5; reasoning effort as labeled, otherwise xhigh per the article footnote; research environment.
在排名中查看此成绩
Mean pass rate33.2%原文 33.2% · 单位 %
公布 未确认
GPT-5.5 Pro
设置
Same underlying model with a setting that uses parallel test-time compute.
未知项
Parallel compute budget and exact Pro harness settings are not stated.
工具
允许工具
推理预算
未记录
框架
Pro harness
协议
GeneBench full suite; separately reported GPT Pro harness; mean unweighted per-problem pass rate over repeated runs; 95% CI in source excerpt
条件
GPT-5.5 Pro; parallel test-time compute; exact compute budget unspecified.; GPT-5.5 Pro uses parallel test-time compute; report separates Pro-harness runs; compute budget unspecified.
在排名中查看此成绩
Mean pass rate1.9%原文 1.9% · 单位 %
公布 未确认
GPT-5.5 none
设置
Interactive chart labels reasoning effort none; chart-specific evaluation settings are otherwise not stated.
未知项
未另记
工具
允许工具
推理预算
none
框架
未记录
协议
GeneBench full suite; 103 evaluations across 10 domains; mean unweighted per-problem pass rate over repeated runs; 95% CI in source excerpt
条件
Reasoning effort none; report states GPT-5.5 xhigh for the mainline run.; GPT-5.5; reasoning effort as labeled, otherwise xhigh per the article footnote; research environment.
在排名中查看此成绩
Mean pass rate3.2%原文 3.2% · 单位 %
公布 未确认
GPT-5.5 low
设置
Interactive chart labels reasoning effort low; chart-specific evaluation settings are otherwise not stated.
未知项
未另记
工具
允许工具
推理预算
low
框架
未记录
协议
GeneBench full suite; 103 evaluations across 10 domains; mean unweighted per-problem pass rate over repeated runs; 95% CI in source excerpt
条件
Reasoning effort low; report states GPT-5.5 xhigh for the mainline run.
在排名中查看此成绩
Mean pass rate9.2%原文 9.2% · 单位 %
公布 未确认
GPT-5.5 medium
设置
Interactive chart labels reasoning effort medium; chart-specific evaluation settings are otherwise not stated.
未知项
未另记
工具
允许工具
推理预算
medium
框架
未记录
协议
GeneBench full suite; 103 evaluations across 10 domains; mean unweighted per-problem pass rate over repeated runs; 95% CI in source excerpt
条件
Reasoning effort medium; report states GPT-5.5 xhigh for the mainline run.; GPT-5.5; reasoning effort as labeled, otherwise xhigh per the article footnote; research environment.
在排名中查看此成绩
Mean pass rate22.2%原文 22.2% · 单位 %
公布 未确认
GPT-5.5 high
设置
Interactive chart labels reasoning effort high; chart-specific evaluation settings are otherwise not stated.
未知项
未另记
工具
允许工具
推理预算
high
框架
未记录
协议
GeneBench full suite; 103 evaluations across 10 domains; mean unweighted per-problem pass rate over repeated runs; 95% CI in source excerpt
条件
Reasoning effort high; report states GPT-5.5 xhigh for the mainline run.; GPT-5.5; reasoning effort as labeled, otherwise xhigh per the article footnote; research environment.
在排名中查看此成绩
Mean pass rate3.1%原文 3.1% · 单位 %
公布 未确认
GPT-5.5 low
设置
Interactive chart labels reasoning effort low; chart-specific evaluation settings are otherwise not stated.
未知项
未另记
工具
允许工具
推理预算
low
框架
未记录
协议
未说明
条件
GPT-5.5; reasoning effort as labeled, otherwise xhigh per the article footnote; research environment.
在排名中查看此成绩

出处与披露

测试版本与指标

  • Mean pass rate · % · 越高越好