GPT-5.5 · HealthBench Professional(525 physician-authored conversations) 3893 characters · 附件证据(gpt55-e-healthbench-professional-length--healthbench-professional)

GPT-5.5 的 HealthBench Professional · 525 physician-authored conversations 官方出处。GPT-5.5 model-level safety evaluation · 3893 characters。见《GPT-5.5 System Card PDF》PDF p.12, paragraph following Table 7, Section 5.1。证据 gpt55-e-healthbench-professional-length--healthbench-professional。

原文与定位

GPT-5.5 · HealthBench Professional(525 physician-authored conversations),GPT-5.5 model-level safety evaluation配置,Mean response length 3893 characters(具体成绩公布时间未确认)。见《GPT-5.5 System Card PDF》PDF p.12, paragraph following Table 7, Section 5.1。 查看成绩条件

In the case of HealthBench Professional, GPT-5.5 was longer (3893 characters vs 3308 characters), and had a higher unadjusted score and a higher length-adjusted score. Table 7 reports GPT-5.5 mean response length as 3818 characters.
官方材料
GPT-5.5 System Card PDF · PDF 第 12 页
定位
附件 · PDF p.12, paragraph following Table 7, Section 5.1
材料版本
原文标注日期 2026-04-23;当前材料版本的公布时间未确认。46-page PDF dated April 23, 2026; includes April 24 API-safeguard update notice and August 19, 2026 protein-binding correction.
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

GPT-5.5 · HealthBench Professional(525 physician-authored conversations),GPT-5.5 model-level safety evaluation配置,Mean response length 3893 characters(具体成绩公布时间未确认)。官方出处 GPT-5.5 System Card PDF(附件,原文标注日期 2026-04-23;当前材料版本的公布时间未确认。46-page PDF dated April 23, 2026; includes April 24 API-safeguard update notice and August 19, 2026 protein-binding correction.)PDF p.12, paragraph following Table 7, Section 5.1,https://deploymentsafety.openai.com/gpt-5-5/gpt-5-5.pdf#page=12。Benchlog https://benchlog.lekshome.cc/evidence/gpt-5-5/gpt55-e-healthbench-professional-length--healthbench-professional/

关联成绩与条件

GPT-5.5 · HealthBench Professional 成绩与配置
指标与成绩配置与评测条件证据关系
Mean response length3893 characters原文 3893 characters · 单位 characters
公布 未确认

结果冲突待核对,精确排名不采用。

GPT-5.5 model-level safety evaluation
设置
GPT-5.5 model results for the production release; each benchmark has its stated prompt, task, and safeguard conditions.
未知项
The card does not state one universal system prompt, reasoning effort, or sampling configuration across evaluations.
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
OpenAI 系统卡 HealthBench 计分;2,000 字符不调整,每多 500 字符减 1.47 分,较短回复对应加分;grader 未说明。
条件
Body text reports 3893 characters, conflicting with Table 7's 3818; no correction notice is provided.

直接支持该成绩