Claude Opus 5.5 · HealthBench(48,000+ rubric items across 5,000 multi-turn conversations) · 附件证据(ev-op55-card-healthbench)

Claude Opus 5.5 的 HealthBench · 48,000+ rubric items across 5,000 multi-turn conversations 官方出处。见《Claude Opus 5.5 System Card》§8.15.1, p.213, Figure 8.15.1.A。证据 ev-op55-card-healthbench。

原文与定位

Claude Opus 5.5 · HealthBench(48,000+ rubric items across 5,000 multi-turn conversations),HealthBench max effort with safeguards配置,Raw score 68.1%(具体成绩公布时间未确认)。见《Claude Opus 5.5 System Card》§8.15.1, p.213, Figure 8.15.1.A。 查看成绩条件

Claude Opus 5.5 · HealthBench(48,000+ rubric items across 5,000 multi-turn conversations),HealthBench max effort with safeguards配置,Length-adjusted score 60.6%(具体成绩公布时间未确认)。见《Claude Opus 5.5 System Card》§8.15.1, p.213, Figure 8.15.1.A。 查看成绩条件

Claude Opus 5.5 achieved a raw score of 68.1% ... After length adjustment ... achieved a score of 60.6%.
官方材料
Claude Opus 5.5 System Card · PDF 第 213 页
定位
附件 · §8.15.1, p.213, Figure 8.15.1.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 · HealthBench(48,000+ rubric items across 5,000 multi-turn conversations),HealthBench max effort with safeguards配置,Raw score 68.1%(具体成绩公布时间未确认);Claude Opus 5.5 · HealthBench(48,000+ rubric items across 5,000 multi-turn conversations),HealthBench max effort with safeguards配置,Length-adjusted score 60.6%(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)§8.15.1, p.213, Figure 8.15.1.A,https://www.anthropic.com/claude-opus-5-5-system-card#page=213。Benchlog https://benchlog.lekshome.cc/evidence/claude-opus-5-5/ev-op55-card-healthbench/

关联成绩与条件

Claude Opus 5.5 · HealthBench 成绩与配置
指标与成绩配置与评测条件证据关系
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.

直接支持该成绩

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.

直接支持该成绩

在排名中查看此成绩