Values by metric family

Each value as the document printed it. Select a value for its source, its checks and its history. Values from other documents are listed after the model’s own, each marked as the first report of that measure or as a restatement.

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

M4 · Misalignment audits 2 values

M4 Misalignment audits: values about GPT-5-Codex
Evaluation and metricValueDocument
From other documents
Destructive action avoidanceAvoidanceGPT-5.1-Codex-Max System Card18 Nov 2025First reportedRestated later: compare with the later valueGPT-5.1-Codex-Max System Card18 Nov 2025First reportedRestated later: compare with the later value
Destructive action avoidanceAvoidanceGPT-5.2-Codex System Card Addendum18 Dec 2025Restated: compare with the earlier valueGPT-5.2-Codex System Card Addendum18 Dec 2025Restated: compare with the earlier value

M7 · Harmful compliance and over-refusal 13 values

M7 Harmful compliance and over-refusal: values about GPT-5-Codex
Evaluation and metricValueDocument
From its own documentGPT-5-Codex System Card Addendum · 15 Sep 2025
Malware refusals (golden set)Refusal rateRestated later: compare with the later valueGPT-5-Codex System Card Addendum15 Sep 2025Restated later: compare with the later value
Production BenchmarksExtremismGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksHarassment/threateningGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksHate/threateningGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksIllicit/non-violentGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksIllicit/violentGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksNon-violent hateGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksPersonal dataGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksSelf-harm/instructionsGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksSelf-harm/intentGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksSexual/exploitativeGPT-5-Codex System Card Addendum15 Sep 2025
Production BenchmarksSexual/minorsGPT-5-Codex System Card Addendum15 Sep 2025
From other documents
Malware refusals (golden set)Refusal rateGPT-5.1-Codex-Max System Card18 Nov 2025Restated: compare with the earlier valueGPT-5.1-Codex-Max System Card18 Nov 2025Restated: compare with the earlier value

M8 · Jailbreak robustness 4 values

M8 Jailbreak robustness: values about GPT-5-Codex
Evaluation and metricValueDocument
From its own documentGPT-5-Codex System Card Addendum · 15 Sep 2025
StrongRejectAbuse/disinformation/hateGPT-5-Codex System Card Addendum15 Sep 2025
StrongRejectIllicit/non-violent crimeGPT-5-Codex System Card Addendum15 Sep 2025
StrongRejectSexual contentGPT-5-Codex System Card Addendum15 Sep 2025
StrongRejectViolenceGPT-5-Codex System Card Addendum15 Sep 2025

M9 · Prompt injection 2 values

M9 Prompt injection: values about GPT-5-Codex
Evaluation and metricValueDocument
From its own documentGPT-5-Codex System Card Addendum · 15 Sep 2025
Prompt injection (Codex env)Attacks successfully ignoredRestated later: compare with the later valueGPT-5-Codex System Card Addendum15 Sep 2025Restated later: compare with the later value
From other documents
Prompt injection (Codex env)Attacks successfully ignoredGPT-5.1-Codex-Max System Card18 Nov 2025Restated: compare with the earlier valueGPT-5.1-Codex-Max System Card18 Nov 2025Restated: compare with the earlier value

M10 · Dangerous capabilities and risk determinations 2 values

M10 Dangerous capabilities and risk determinations: values about GPT-5-Codex
Evaluation and metricValueDocument
From its own documentGPT-5-Codex System Card Addendum · 15 Sep 2025
Preparedness Framework determinationBiological and chemicalGPT-5-Codex System Card Addendum15 Sep 2025
Preparedness Framework determinationCybersecurityGPT-5-Codex System Card Addendum15 Sep 2025

Restated in later documents

A later document reported values about GPT-5-Codex again. Each pair is shown side by side: both values are kept with their own documents, and the later one does not replace the earlier.

Values about GPT-5-Codex restated in later documents
Earlier valueLater value
M4 · Misalignment audits
Destructive action avoidanceAvoidance
GPT-5.1-Codex-Max System Card18 Nov 2025 · first reportedGPT-5.2-Codex System Card Addendum18 Dec 2025No reason stated
M7 · Harmful compliance and over-refusal
Malware refusals (golden set)Refusal rate
GPT-5-Codex System Card Addendum15 Sep 2025 · its own documentGPT-5.1-Codex-Max System Card18 Nov 2025No reason stated
M9 · Prompt injection
Prompt injection (Codex env)Attacks successfully ignored
GPT-5-Codex System Card Addendum15 Sep 2025 · its own documentGPT-5.1-Codex-Max System Card18 Nov 2025No reason stated

Risk determinations

The developer's formal decisions about GPT-5-Codex under its framework, as printed. Levels from different frameworks do not map onto one another.

Risk determinations about GPT-5-Codex
DomainLevel as printedFramework and document
Bio/chemHigh (treated as) Preparedness Framework GPT-5-Codex System Card Addendum · 15 Sep 2025Preparedness Framework GPT-5-Codex System Card Addendum · 15 Sep 2025
CyberBelow High Preparedness Framework GPT-5-Codex System Card Addendum · 15 Sep 2025Preparedness Framework GPT-5-Codex System Card Addendum · 15 Sep 2025