Home Model series GPT Sol GPT-5.6 Sol GPT-5.6 Sol · HealthBench · 48,000+ rubric items across 5,000 multi-turn conversations · Scores and configurations GPT-5.6 Sol · HealthBench · 48,000+ rubric items across 5,000 multi-turn conversations · Scores and configurations GPT-5.6 Sol record for HealthBench · 48,000+ rubric items across 5,000 multi-turn conversations. 3 score records for this release, including 3 exact values (may include historical records). Includes 2 source locations.
Scores and evaluation conditions GPT-5.6 Sol · HealthBench scores and configurations Metric and score Configuration and evaluation conditions Sources 长度校正分(0–100) 57 score Reported 57.0 · Unit score Published Unconfirmed
GPT-5.6 Sol HealthBench evaluations Settings HealthBench and HealthBench Professional-family evaluations; table reports length-adjusted scores. Unknowns Reasoning effort、Grader version for each GPT-5.6 score、System-level safeguards、Exact benchmark data snapshot Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol OpenAI 系统卡 HealthBench 计分;2,000 字符不调整,每多 500 字符减 2.99 分,较短回复对应加分;grader 未说明。 Conditions Length penalty is 2.99 points per 500 characters beyond 2,000; response length 1,764 characters. 系统卡明确采用 0–100 量表;保留原始分数字符串,不乘除数值。 View this score in rankings 未校正分(0–100) 55.6 score Reported 55.6 · Unit score Published Unconfirmed
GPT-5.6 Sol HealthBench evaluations Settings HealthBench and HealthBench Professional-family evaluations; table reports length-adjusted scores. Unknowns Reasoning effort、Grader version for each GPT-5.6 score、System-level safeguards、Exact benchmark data snapshot Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol OpenAI 系统卡 HealthBench 计分;2,000 字符不调整,每多 500 字符减 2.99 分,较短回复对应加分;grader 未说明。 Conditions Table reports the tuple as length-adjusted score (unadjusted score, mean response length in characters). 系统卡明确采用 0–100 量表;保留原始分数字符串,不乘除数值。 View this score in rankings Mean response length 1764 characters Reported 1764 · Unit characters Published Unconfirmed
GPT-5.6 Sol HealthBench evaluations Settings HealthBench and HealthBench Professional-family evaluations; table reports length-adjusted scores. Unknowns Reasoning effort、Grader version for each GPT-5.6 score、System-level safeguards、Exact benchmark data snapshot Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol OpenAI 系统卡 HealthBench 计分;2,000 字符不调整,每多 500 字符减 2.99 分,较短回复对应加分;grader 未说明。 Conditions Table reports the tuple as length-adjusted score (unadjusted score, mean response length in characters). View this score in rankings
Benchmark versions and metrics Aliases HealthBench
Length-adjusted score · % · Higher is better Raw score · % · Higher is better Mean response length · characters · Ranking direction unconfirmed 长度校正分(0–100) · score · Higher is better 未校正分(0–100) · score · Higher is better 相对 GPT-5.4 的长度校正分差 · points · Ranking direction unconfirmed Unadjusted score · % · Higher is better