Claude Haiku 5.5 · HealthBench Professional · 525 physician-authored conversations · 成绩与配置

Claude Haiku 5.5 的 HealthBench Professional · 525 physician-authored conversations 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 1 条来源定位。

成绩与评测条件

Claude Haiku 5.5 · HealthBench Professional 成绩与配置
指标与成绩配置与评测条件出处
Length-adjusted score57.9%原文 57.9% · 单位 %
公布 未确认
Healthcare evaluations / low
设置
Safety classifiers enabled; low effort; run counts, scoring variants, and dates are recorded per result.
未知项
Healthcare evaluations use benchmark-specific run counts and scoring variants.
工具
禁止工具
推理预算
low
框架
未记录
协议
HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
条件
low effort. Low through xhigh runs were made October 4; max runs were made October 1-2. The card also reports 13 max-effort repeat runs on October 4 averaging 64.2%; kept separate from the primary 64.8% result. Labeled figure value.
在排名中查看此成绩
Length-adjusted score59.9%原文 59.9% · 单位 %
公布 未确认
Healthcare evaluations / medium
设置
Safety classifiers enabled; medium effort; run counts, scoring variants, and dates are recorded per result.
未知项
Healthcare evaluations use benchmark-specific run counts and scoring variants.
工具
禁止工具
推理预算
medium
框架
未记录
协议
HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
条件
medium effort. Low through xhigh runs were made October 4; max runs were made October 1-2. The card also reports 13 max-effort repeat runs on October 4 averaging 64.2%; kept separate from the primary 64.8% result. Labeled figure value.
在排名中查看此成绩
Length-adjusted score61.3%原文 61.3% · 单位 %
公布 未确认
Healthcare evaluations / high
设置
Safety classifiers enabled; high effort; run counts, scoring variants, and dates are recorded per result.
未知项
Healthcare evaluations use benchmark-specific run counts and scoring variants.
工具
禁止工具
推理预算
high
框架
未记录
协议
HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
条件
high effort. Low through xhigh runs were made October 4; max runs were made October 1-2. The card also reports 13 max-effort repeat runs on October 4 averaging 64.2%; kept separate from the primary 64.8% result. Labeled figure value.
在排名中查看此成绩
Length-adjusted score61%原文 61.0% · 单位 %
公布 未确认
Healthcare evaluations / xhigh
设置
Safety classifiers enabled; xhigh effort; run counts, scoring variants, and dates are recorded per result.
未知项
Healthcare evaluations use benchmark-specific run counts and scoring variants.
工具
禁止工具
推理预算
xhigh
框架
未记录
协议
HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
条件
xhigh effort. Low through xhigh runs were made October 4; max runs were made October 1-2. The card also reports 13 max-effort repeat runs on October 4 averaging 64.2%; kept separate from the primary 64.8% result. Labeled figure value.
在排名中查看此成绩
Length-adjusted score64.8%原文 64.8% · 单位 %
公布 未确认
Healthcare evaluations / max
设置
Safety classifiers enabled; max effort; run counts, scoring variants, and dates are recorded per result.
未知项
Healthcare evaluations use benchmark-specific run counts and scoring variants.
工具
禁止工具
推理预算
max
框架
未记录
协议
HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
条件
max effort. Low through xhigh runs were made October 4; max runs were made October 1-2. The card also reports 13 max-effort repeat runs on October 4 averaging 64.2%; kept separate from the primary 64.8% result. Labeled figure value.
在排名中查看此成绩
Raw score61.6%原文 61.6% raw · 单位 %
公布 未确认
Healthcare evaluations / low
设置
Safety classifiers enabled; low effort; run counts, scoring variants, and dates are recorded per result.
未知项
Healthcare evaluations use benchmark-specific run counts and scoring variants.
工具
工具条件未记录
推理预算
low
框架
未记录
协议
HealthBench Professional raw score before length adjustment.
条件
Low effort; five runs; October 4.
在排名中查看此成绩
Raw score71%原文 71.0% raw · 单位 %
公布 未确认
Healthcare evaluations / max
设置
Safety classifiers enabled; max effort; run counts, scoring variants, and dates are recorded per result.
未知项
Healthcare evaluations use benchmark-specific run counts and scoring variants.
工具
工具条件未记录
推理预算
max
框架
未记录
协议
HealthBench Professional raw score before length adjustment.
条件
Max effort; five runs, October 1-2. Separate 13-run max repeat average is reported as 64.2% length-adjusted and is not substituted here.
在排名中查看此成绩

出处与披露

测试版本与指标

别名 HealthBench Professional

  • Raw score · % · 越高越好
  • Length-adjusted score · % · 越高越好
  • Score · % · 越高越好
  • Mean response length · characters · 排序方向未确认
  • 长度校正分(0–100) · score · 越高越好
  • 未校正分(0–100) · score · 越高越好
  • 相对 GPT-5.4 的长度校正分差 · points · 排序方向未确认
  • Unadjusted score · % · 越高越好