Claude Opus · Protocol Understanding · V2; Benchling-created evaluation using real laboratory protocols and applied questions; task count unspecified · 历代记录

按发布时间查看 Claude Opus 的 2 次发布在 Protocol Understanding · V2; Benchling-created evaluation using real laboratory protocols and applied questions; task count unspecified 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 1 条成绩记录,含 1 条精确值(可含历史记录)

Claude Opus 5 · Protocol Understanding 成绩与配置
指标与成绩配置与评测条件出处
Score78.4%原文 78.4% · 单位 %
公布 未确认
Protocols · max
设置
Card-wide default max effort; bash, file editor, and web search tools; separate Troubleshooting and Benchling Understanding V2 variants.
未知项
未另记
工具
允许工具
推理预算
max
框架
未记录
协议
Benchling-created Protocols Understanding V2 variant.
条件
Card-wide default max effort; bash, file editor and web search tools.
在排名中查看此成绩

Claude Opus 5.5

· 本次发布有 1 条成绩记录,含 1 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 · Protocol Understanding 成绩与配置
指标与成绩配置与评测条件出处
Score69%原文 69.0% · 单位 %
公布 未确认
System Card max effort (per-result benchmark harnesses)
设置
System Card §8.1 default: adaptive thinking/max effort and default sampling settings unless a section says otherwise. Benchmark tools, evaluator, safeguards, fallback, and repetitions remain per-result conditions; this config represents the stated model effort, not harness equivalence.
未知项
Exact run time is not disclosed.、Per-benchmark tools, safeguards, and repetition counts are specified only where the source states them.
工具
允许工具
推理预算
max
框架
未记录
协议
Protocols Understanding V2; 119 Benchling-created held-out problems; bash, file editor and web-search tools; five attempts per problem unless otherwise stated.
条件
Named domain-capability result; separately cross-referenced in the CB risk assessment, but not a CB score or a safeguarding intervention. Source-wide conditions: Troubleshooting: 206 problems. Understanding V2: 119 Benchling-created held-out problems. Both variants have bash, file editor and web-search tools. Also cross-referenced by the card's CB risk assessment; retain as named domain-capability metrics and do not convert them into a CB risk score.
在排名中查看此成绩