Values by metric family

Each value as the document printed it. Select a value for its source, its checks and its history.

KeyVerifiedUnverifiedDisputedCorrected read off a figure or stated in wordsA value opens its source and history.

M5 · Honesty and hallucination 3 values

M5 Honesty and hallucination: values about gpt-5-main
Evaluation and metricValueDocument
From its own documentGPT-5 System Card · 7 Aug 2025
Production-traffic factualityClaim-level hallucination rate reduction vs GPT-4oCondition: with browsing; production promptsGPT-5 System Card7 Aug 2025
Production-traffic factualityReduction in responses with 1+ major error vs GPT-4oCondition: with browsing; production promptsGPT-5 System Card7 Aug 2025
SimpleQAHallucination rateCondition: no browsingGPT-5 System Card7 Aug 2025

M6 · Sycophancy 3 values

M6 Sycophancy: values about gpt-5-main
Evaluation and metricValueDocument
From its own documentGPT-5 System Card · 7 Aug 2025
Sycophancy offline evalSycophancy scoreCondition: offlineGPT-5 System Card7 Aug 2025
Sycophancy online prevalence (A/B)Relative change vs GPT-4oCondition: free usersGPT-5 System Card7 Aug 2025
Sycophancy online prevalence (A/B)Relative change vs GPT-4oCondition: paid usersGPT-5 System Card7 Aug 2025

M7 · Harmful compliance and over-refusal 11 values

M7 Harmful compliance and over-refusal: values about gpt-5-main
Evaluation and metricValueDocument
From its own documentGPT-5 System Card · 7 Aug 2025
Production BenchmarksExtremismGPT-5 System Card7 Aug 2025
Production BenchmarksHarassment/threateningGPT-5 System Card7 Aug 2025
Production BenchmarksHate/threateningGPT-5 System Card7 Aug 2025
Production BenchmarksIllicit/non-violentGPT-5 System Card7 Aug 2025
Production BenchmarksIllicit/violentGPT-5 System Card7 Aug 2025
Production BenchmarksNon-violent hateGPT-5 System Card7 Aug 2025
Production BenchmarksPersonal dataGPT-5 System Card7 Aug 2025
Production BenchmarksSelf-harm/instructionsGPT-5 System Card7 Aug 2025
Production BenchmarksSelf-harm/intentGPT-5 System Card7 Aug 2025
Production BenchmarksSexual/exploitativeGPT-5 System Card7 Aug 2025
Production BenchmarksSexual/minorsGPT-5 System Card7 Aug 2025

M8 · Jailbreak robustness 4 values

M8 Jailbreak robustness: values about gpt-5-main
Evaluation and metricValueDocument
From its own documentGPT-5 System Card · 7 Aug 2025
StrongRejectAbuse/disinformation/hateGPT-5 System Card7 Aug 2025
StrongRejectIllicit/non-violent crimeGPT-5 System Card7 Aug 2025
StrongRejectSexual contentGPT-5 System Card7 Aug 2025
StrongRejectViolenceGPT-5 System Card7 Aug 2025