GPT-6.2 Sol · HealthBench Professional(525 physician-authored conversations) · 附件证据(ev-gpt62-sol-healthbench-professional-table17)

GPT-6.2 Sol 的 HealthBench Professional · 525 physician-authored conversations 官方出处。见《GPT-6.2 Astra and Sol System Card (PDF)》PDF physical page 74 / printed page 73 / section 3.4.1 / Table 17 / GPT-6.2 Sol column。证据 ev-gpt62-sol-healthbench-professional-table17。

原文与定位

GPT-6.2 Sol · HealthBench Professional(525 physician-authored conversations),GPT-6.2 Sol, HealthBench Professional configuration unspecified配置,长度校正分(0–100) 65.9 score(具体成绩公布时间未确认)。见《GPT-6.2 Astra and Sol System Card (PDF)》PDF physical page 74 / printed page 73 / section 3.4.1 / Table 17 / GPT-6.2 Sol column。 查看成绩条件

GPT-6.2 Sol · HealthBench Professional(525 physician-authored conversations),GPT-6.2 Sol, HealthBench Professional configuration unspecified配置,未校正分(0–100) 65.5 score(具体成绩公布时间未确认)。见《GPT-6.2 Astra and Sol System Card (PDF)》PDF physical page 74 / printed page 73 / section 3.4.1 / Table 17 / GPT-6.2 Sol column。 查看成绩条件

GPT-6.2 Sol · HealthBench Professional(525 physician-authored conversations),GPT-6.2 Sol, HealthBench Professional configuration unspecified配置,Mean response length 1869 characters(具体成绩公布时间未确认)。见《GPT-6.2 Astra and Sol System Card (PDF)》PDF physical page 74 / printed page 73 / section 3.4.1 / Table 17 / GPT-6.2 Sol column。 查看成绩条件

GPT-6.2 Astra and GPT-6.2 Sol scored 66.3 and 65.9 after adjustment, compared with 64.7 for GPT-6 Astra. Their raw scores were 66.0 and 65.5 versus 68.2, and their responses averaged 1,884 and 1,869 characters versus 3,185. Table 17: HealthBench Professional evaluations, reported as length-adjusted score (unadjusted score, mean visible response length in characters). Scores are on a 0–100 scale; higher is better.
官方材料
GPT-6.2 Astra and Sol System Card (PDF) · PDF 第 74 页
定位
附件 · PDF physical page 74 / printed page 73 / section 3.4.1 / Table 17 / GPT-6.2 Sol column
材料版本
官方HTML及119页PDF,封面和change log日期2026-11-17;当前页面后续修订时间未知。119页全文(含脚注和参考文献)已逐页核读,21张表和56张图均已视觉核对;HTML全部119个section与PDF正文及表图对应已核对。安全、Preparedness、对齐与monitorability专项已按用途登记,不自动并入常规能力展示。
版本公布
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-6.2 Sol · HealthBench Professional(525 physician-authored conversations),GPT-6.2 Sol, HealthBench Professional configuration unspecified配置,长度校正分(0–100) 65.9 score(具体成绩公布时间未确认);GPT-6.2 Sol · HealthBench Professional(525 physician-authored conversations),GPT-6.2 Sol, HealthBench Professional configuration unspecified配置,未校正分(0–100) 65.5 score(具体成绩公布时间未确认);GPT-6.2 Sol · HealthBench Professional(525 physician-authored conversations),GPT-6.2 Sol, HealthBench Professional configuration unspecified配置,Mean response length 1869 characters(具体成绩公布时间未确认)。官方出处 GPT-6.2 Astra and Sol System Card (PDF)(附件,官方HTML及119页PDF,封面和change log日期2026-11-17;当前页面后续修订时间未知。119页全文(含脚注和参考文献)已逐页核读,21张表和56张图均已视觉核对;HTML全部119个section与PDF正文及表图对应已核对。安全、Preparedness、对齐与monitorability专项已按用途登记,不自动并入常规能力展示。)PDF physical page 74 / printed page 73 / section 3.4.1 / Table 17 / GPT-6.2 Sol column,https://deploymentsafety.openai.com/gpt-6-2-astra-sol/gpt-6-2-astra-sol.pdf#page=74。Benchlog https://benchlog.lekshome.cc/evidence/gpt-6-2-sol/ev-gpt62-sol-healthbench-professional-table17/

关联成绩与条件

GPT-6.2 Sol · HealthBench Professional 成绩与配置
指标与成绩配置与评测条件证据关系
长度校正分(0–100)65.9 score原文 65.9 · 单位 score
公布 未确认
GPT-6.2 Sol, HealthBench Professional configuration unspecified
设置
GPT-6.2 Sol; Table 17 does not state reasoning effort. Length adjustment: 1.47 points per 500 characters above/below 2,000; models are not told the adjustment.
未知项
HealthBench Professional 推理档位未注明、采样、完整 harness 和工具配置未注明
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).
条件
分数为0–100尺度,不是百分比。长度校正以2000字符为基准,每500字符对应1.47分的惩罚或奖励;评测时不向模型告知该校正规则。推理档位及逐项采样设置未注明。

直接支持该成绩

在排名中查看此成绩
未校正分(0–100)65.5 score原文 65.5 · 单位 score
公布 未确认
GPT-6.2 Sol, HealthBench Professional configuration unspecified
设置
GPT-6.2 Sol; Table 17 does not state reasoning effort. Length adjustment: 1.47 points per 500 characters above/below 2,000; models are not told the adjustment.
未知项
HealthBench Professional 推理档位未注明、采样、完整 harness 和工具配置未注明
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).
条件
分数为0–100尺度,不是百分比。长度校正以2000字符为基准,每500字符对应1.47分的惩罚或奖励;评测时不向模型告知该校正规则。推理档位及逐项采样设置未注明。

直接支持该成绩

在排名中查看此成绩
Mean response length1869 characters原文 1,869 · 单位 characters
公布 未确认
GPT-6.2 Sol, HealthBench Professional configuration unspecified
设置
GPT-6.2 Sol; Table 17 does not state reasoning effort. Length adjustment: 1.47 points per 500 characters above/below 2,000; models are not told the adjustment.
未知项
HealthBench Professional 推理档位未注明、采样、完整 harness 和工具配置未注明
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
Table 17: length-adjusted score (unadjusted score, mean visible response length in characters).
条件
分数为0–100尺度,不是百分比。长度校正以2000字符为基准,每500字符对应1.47分的惩罚或奖励;评测时不向模型告知该校正规则。推理档位及逐项采样设置未注明。

直接支持该成绩

在排名中查看此成绩