Claude Opus · HealthBench · 48,000+ rubric items across 5,000 multi-turn conversations · 历代记录

按发布时间查看 Claude Opus 的 2 次发布在 HealthBench · 48,000+ rubric items across 5,000 multi-turn conversations 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 2 条成绩记录,含 2 条精确值(可含历史记录)

Claude Opus 5 · HealthBench 成绩与配置
指标与成绩配置与评测条件出处
Raw score67.1%原文 67.1% · 单位 %
公布 未确认
HealthBench · max
设置
Max effort; no tools or customized system prompts; Opus 4.8 grader; five-trial average; length-adjustment method cited from the GPT-5.5 system card.
未知项
未另记
工具
禁止工具
推理预算
max
框架
未记录
协议
HealthBench; raw rubric score.
条件
Max effort; no tools or customized system prompt; Opus 4.8 grader; five-trial average.
在排名中查看此成绩
Length-adjusted score57.8%原文 57.8% · 单位 %
公布 未确认
HealthBench · max
设置
Max effort; no tools or customized system prompts; Opus 4.8 grader; five-trial average; length-adjustment method cited from the GPT-5.5 system card.
未知项
未另记
工具
禁止工具
推理预算
max
框架
未记录
协议
HealthBench; length-adjusted score.
条件
Max effort; no tools or customized system prompt; Opus 4.8 grader; five-trial average.
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 · HealthBench 成绩与配置
指标与成绩配置与评测条件出处
Raw score68.1%原文 68.1% raw rubric score · 单位 %
公布 未确认
HealthBench max effort with safeguards
设置
Adaptive thinking/max effort; safety classifiers enabled and refusal fallback to Opus 5; no tools or customized system prompts; five trials.
未知项
未另记
工具
禁止工具
推理预算
max
框架
未记录
协议
HealthBench; raw score; five trials; no tools or customized system prompt.
条件
Safety classifiers enabled with refusal fallback to Opus 5. Source-wide conditions: Five trials; no tools or customized system prompts; safety classifiers enabled with refusal fallback to Opus 5.
在排名中查看此成绩
Length-adjusted score60.6%原文 60.6% length-adjusted score · 单位 %
公布 未确认
HealthBench max effort with safeguards
设置
Adaptive thinking/max effort; safety classifiers enabled and refusal fallback to Opus 5; no tools or customized system prompts; five trials.
未知项
未另记
工具
禁止工具
推理预算
max
框架
未记录
协议
HealthBench; length-adjusted score; five trials; no tools or customized system prompt.
条件
Length adjustment penalizes verbose responses; safety classifiers enabled with refusal fallback to Opus 5. Source-wide conditions: Five trials; no tools or customized system prompts; safety classifiers enabled with refusal fallback to Opus 5.
在排名中查看此成绩