HealthBench Professional · 525 physician-authored conversations · 测试定义与成绩

HealthBench Professional · 525 physician-authored conversations 收录 21 次模型发布的评测披露或成绩。

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

各系列最新模型 · 公布成绩

常规展示范围。排名按指标与评测协议分组,已记录的工具、预算及框架分别展示。条件未说明的组保留公布值供查证。

Raw score %

工具条件未记录 / max / 框架未记录 · HealthBench Professional; Artificial Analysis; GPT-5.4 low-effort grader; max effort.

HealthBench Professional · 525 physician-authored conversations · Raw score · 工具条件未记录 / max / 框架未记录 · HealthBench Professional; Artificial Analysis; GPT-5.4 low-effort grader; max effort.
模型与配置成绩(%)记录与证据
Claude Fable 5.2Max71.1%

工具条件未记录 / low / 框架未记录 · HealthBench Professional raw score before length adjustment.

HealthBench Professional · 525 physician-authored conversations · Raw score · 工具条件未记录 / low / 框架未记录 · HealthBench Professional raw score before length adjustment.
模型与配置成绩(%)记录与证据
Claude Haiku 5.5Low61.6%

工具条件未记录 / max / 框架未记录 · HealthBench Professional raw score before length adjustment.

HealthBench Professional · 525 physician-authored conversations · Raw score · 工具条件未记录 / max / 框架未记录 · HealthBench Professional raw score before length adjustment.
模型与配置成绩(%)记录与证据
Claude Haiku 5.5Max71%

工具条件未记录 / max / 框架未记录 · HealthBench Professional, max effort.

HealthBench Professional · 525 physician-authored conversations · Raw score · 工具条件未记录 / max / 框架未记录 · HealthBench Professional, max effort.
模型与配置成绩(%)记录与证据
Claude Sonnet 5.5Max77.1%

禁止工具 / max / 框架未记录 · HealthBench Professional; raw score; five trials; no tools or customized system prompt.

HealthBench Professional · 525 physician-authored conversations · Raw score · 禁止工具 / max / 框架未记录 · HealthBench Professional; raw score; five trials; no tools or customized system prompt.
模型与配置成绩(%)记录与证据
Claude Opus 5.5Max77.1%

Length-adjusted score %

工具条件未记录 / max / 框架未记录 · HealthBench Professional length-adjusted score; Artificial Analysis; GPT-5.4 low-effort grader; max effort.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 工具条件未记录 / max / 框架未记录 · HealthBench Professional length-adjusted score; Artificial Analysis; GPT-5.4 low-effort grader; max effort.
模型与配置成绩(%)记录与证据
Claude Fable 5.2Max67.7%

禁止工具 / low / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 禁止工具 / low / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
模型与配置成绩(%)记录与证据
Claude Haiku 5.5Low57.9%

禁止工具 / medium / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 禁止工具 / medium / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
模型与配置成绩(%)记录与证据
Claude Haiku 5.5Medium59.9%

禁止工具 / high / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 禁止工具 / high / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
模型与配置成绩(%)记录与证据
Claude Haiku 5.5High61.3%

禁止工具 / xhigh / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 禁止工具 / xhigh / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
模型与配置成绩(%)记录与证据
Claude Haiku 5.5Xhigh61%

禁止工具 / max / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 禁止工具 / max / 框架未记录 · HealthBench Professional length-adjusted score; Claude Opus 4.8 grader; no tools or customized system prompt; five runs per effort level.
模型与配置成绩(%)记录与证据
Claude Haiku 5.5Max64.8%

工具条件未记录 / max / 框架未记录 · HealthBench Professional, max effort.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 工具条件未记录 / max / 框架未记录 · HealthBench Professional, max effort.
模型与配置成绩(%)记录与证据
Claude Sonnet 5.5Max69.2%

工具条件未记录 / low / 框架未记录 · HealthBench Professional length-adjusted score.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 工具条件未记录 / low / 框架未记录 · HealthBench Professional length-adjusted score.
模型与配置成绩(%)记录与证据
Claude Sonnet 5.5Low67.9%

工具条件未记录 / medium / 框架未记录 · HealthBench Professional length-adjusted score.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 工具条件未记录 / medium / 框架未记录 · HealthBench Professional length-adjusted score.
模型与配置成绩(%)记录与证据
Claude Sonnet 5.5Medium67.6%

工具条件未记录 / high / 框架未记录 · HealthBench Professional length-adjusted score.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 工具条件未记录 / high / 框架未记录 · HealthBench Professional length-adjusted score.
模型与配置成绩(%)记录与证据
Claude Sonnet 5.5High67.8%

工具条件未记录 / xhigh / 框架未记录 · HealthBench Professional length-adjusted score.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 工具条件未记录 / xhigh / 框架未记录 · HealthBench Professional length-adjusted score.
模型与配置成绩(%)记录与证据
Claude Sonnet 5.5Xhigh68.2%

禁止工具 / max / 框架未记录 · HealthBench Professional; length-adjusted score; five trials; no tools or customized system prompt.

HealthBench Professional · 525 physician-authored conversations · Length-adjusted score · 禁止工具 / max / 框架未记录 · HealthBench Professional; length-adjusted score; five trials; no tools or customized system prompt.
模型与配置成绩(%)记录与证据
Claude Opus 5.5Max65.6%

Mean response length characters

Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).

HealthBench Professional · 525 physician-authored conversations · Mean response length · Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).
模型与配置成绩(characters)记录与证据
GPT-6.2 Astra未注明1884 characters
GPT-6.2 Sol未注明1869 characters

System-card HealthBench table: length-adjusted score (unadjusted score, mean response length in characters).

HealthBench Professional · 525 physician-authored conversations · Mean response length · System-card HealthBench table: length-adjusted score (unadjusted score, mean response length in characters).
模型与配置成绩(characters)记录与证据
GPT-6.1 Terra未注明2531 characters

System-card Table 7: mean response length paired with adjusted and unadjusted scores.

HealthBench Professional · 525 physician-authored conversations · Mean response length · System-card Table 7: mean response length paired with adjusted and unadjusted scores.
模型与配置成绩(characters)记录与证据
GPT-6 Luna未注明2119 characters

长度校正分(0–100) score

Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).

HealthBench Professional · 525 physician-authored conversations · 长度校正分(0–100) · Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).
名次模型与配置成绩(score)记录与证据
1GPT-6.2 Astra未注明66.3 score
2GPT-6.2 Sol未注明65.9 score

System-card HealthBench table: length-adjusted score (unadjusted score, mean response length in characters).

HealthBench Professional · 525 physician-authored conversations · 长度校正分(0–100) · System-card HealthBench table: length-adjusted score (unadjusted score, mean response length in characters).
模型与配置成绩(score)记录与证据
GPT-6.1 Terra未注明62.7 score

System-card Table 7: length-adjusted score (unadjusted score, mean response length in characters).

HealthBench Professional · 525 physician-authored conversations · 长度校正分(0–100) · System-card Table 7: length-adjusted score (unadjusted score, mean response length in characters).
模型与配置成绩(score)记录与证据
GPT-6 Luna未注明60.8 score

未校正分(0–100) score

Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).

HealthBench Professional · 525 physician-authored conversations · 未校正分(0–100) · Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).
名次模型与配置成绩(score)记录与证据
1GPT-6.2 Astra未注明66 score
2GPT-6.2 Sol未注明65.5 score

System-card HealthBench table: length-adjusted score (unadjusted score, mean response length in characters).

HealthBench Professional · 525 physician-authored conversations · 未校正分(0–100) · System-card HealthBench table: length-adjusted score (unadjusted score, mean response length in characters).
模型与配置成绩(score)记录与证据
GPT-6.1 Terra未注明64.2 score

System-card Table 7: unadjusted score paired with the length-adjusted score.

HealthBench Professional · 525 physician-authored conversations · 未校正分(0–100) · System-card Table 7: unadjusted score paired with the length-adjusted score.
模型与配置成绩(score)记录与证据
GPT-6 Luna未注明61.2 score

历史记录 · 21 次发布

  • GPT-6.2 Astra本次发布有 3 条成绩记录,含 3 条精确值(可含历史记录)
  • GPT-6.2 Sol本次发布有 3 条成绩记录,含 3 条精确值(可含历史记录)
  • Claude Fable 5.2本次发布有 2 条成绩记录,含 2 条精确值(可含历史记录)
  • Gemini 4 Carbon本次发布有 4 条成绩记录,含 4 条精确值(可含历史记录)
  • GPT-6.1 Terra本次发布有 3 条成绩记录,含 3 条精确值(可含历史记录)
  • Claude Haiku 5.5本次发布有 7 条成绩记录,含 7 条精确值(可含历史记录)
  • Gemini 4 Argon本次发布有 4 条成绩记录,含 4 条精确值(可含历史记录)
  • GPT-6.1 Sol本次发布有 3 条成绩记录,含 3 条精确值(可含历史记录)
  • Claude Sonnet 5.5本次发布有 6 条成绩记录,含 6 条精确值(可含历史记录)
  • 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)
  • GPT-6 Luna本次发布有 3 条成绩记录,含 3 条精确值(可含历史记录)
  • GPT-6 Sol本次发布有 3 条成绩记录,含 3 条精确值(可含历史记录)
  • Claude Fable 5.1本次发布有 2 条成绩记录,含 2 条精确值(可含历史记录)
  • GPT-6 Astra本次发布有 9 条成绩记录,含 9 条精确值(可含历史记录)
  • 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)
  • GPT-5.6 Sol本次发布有 4 条成绩记录,含 4 条精确值(可含历史记录)
  • GPT-5.6 Luna本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
  • GPT-5.6 Terra本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
  • Claude Sonnet 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 4.8本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
  • GPT-5.5本次发布有 5 条成绩记录,含 5 条精确值(可含历史记录)