Versions

One version is on record: the copy we retrieved on 26 Sep 2026. We know of no other.

  1. 26 Sep 2026

    Date retrieved

    Copy retrieved 26 Sep 2026

    Date from
    the date we retrieved it; the copy states no version date
    Copy
    Copy retrieved on 26 Sep 2026
    Values
    27 values recorded from this version

No version has a file hash or an archived snapshot yet. From dataset v0.2 each retrieved version carries both (Methodology §7).

Revisions

No revisions are recorded for this document. We know of only one version of it.

Values

Every value we recorded from this document, grouped by metric family and ordered by where the document prints it. Location is the section, table or page as the document numbers it. 4 of the 27 have been blind-verified: a second reader found the same value without seeing ours.

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

M1 Evaluation awareness

1 value

M1 Evaluation awareness: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.4 ThinkingEvaluation awarenessSamples with verbalized eval awarenessRun by Apollo ResearchNone statedExternal (Apollo)Verified

M3 Sabotage and sandbagging

2 values

M3 Sabotage and sandbagging: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.4 ThinkingSabotage capabilityMean sabotage scoreRun by Apollo ResearchNone statedExternal (Apollo)Unverified
GPT-5.4 ThinkingSandbaggingAccuracy drop when incentivized to underperformRun by Apollo ResearchNone statedExternal (Apollo)Unverified

M4 Misalignment audits

2 values

M4 Misalignment audits: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.4 ThinkingCovert deceptive behaviorOverall deception rateRun by Apollo Researchno goal nudgeExternal (Apollo)Unverified
GPT-5.4 ThinkingDestructive action avoidanceAvoidanceNone statedTable 7Verified

M7 Harmful compliance and over-refusal

9 values

M7 Harmful compliance and over-refusal: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.4 ThinkingProduction BenchmarksExtremismNone statedTable 1Unverified
GPT-5.4 ThinkingProduction BenchmarksHarassmentNone statedTable 1Unverified
GPT-5.4 ThinkingProduction BenchmarksHateNone statedTable 1Unverified
GPT-5.4 ThinkingProduction BenchmarksNonviolent illicit behaviorNone statedTable 1Unverified
GPT-5.4 ThinkingProduction BenchmarksSelf-harm (standard)None statedTable 1Unverified
GPT-5.4 ThinkingProduction BenchmarksSexualNone statedTable 1Verified
GPT-5.4 ThinkingProduction BenchmarksSexual/minorsNone statedTable 1Unverified
GPT-5.4 ThinkingProduction BenchmarksViolenceNone statedTable 1Unverified
GPT-5.4 ThinkingProduction BenchmarksViolent illicit behaviorNone statedTable 1Unverified

M8 Jailbreak robustness

1 value

M8 Jailbreak robustness: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.4 ThinkingJailbreaksWorst-case defender successNone statedFig 2Unverified

M9 Prompt injection

2 values

M9 Prompt injection: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.4 ThinkingPrompt injectionConnectorsNone statedTable 4Unverified
GPT-5.4 ThinkingPrompt injectionFunction callsNone statedTable 4Unverified

M10 Dangerous capabilities and risk determinations

7 values

M10 Dangerous capabilities and risk determinations: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.2 ThinkingCyber RangeCombined pass rateNone statedCyber RangeUnverified
GPT-5.2-CodexCyber RangeCombined pass rateNone statedCyber RangeVerified
GPT-5.3-CodexCyber RangeCombined pass rateNone statedCyber RangeUnverified
GPT-5.4 ThinkingCyber RangeCombined pass rateNone statedCyber RangeUnverified
GPT-5.4 ThinkingPreparedness Framework determinationAI self-improvementNone statedPreparednessUnverified
GPT-5.4 ThinkingPreparedness Framework determinationBiological and chemicalNone statedPreparednessUnverified
GPT-5.4 ThinkingPreparedness Framework determinationCybersecurityNone statedPreparednessUnverified

M12 Chain-of-thought monitorability

3 values

M12 Chain-of-thought monitorability: values in the GPT-5.4 Thinking System Card
ModelEvaluationConditionValueLocationChecked
GPT-5.4 ThinkingCoT controllabilityCoTs successfully controlled10k-character CoTsCoTUnverified
GPT-5.4 ThinkingCoT monitorabilityMonitorability vs GPT-5 ThinkingNone statedCoT (Fig 6)Unverified
GPT-5.4 ThinkingMultilingual reasoning anomaliesSamples with anomaliesRun by Apollo ResearchNone statedExternal (Apollo)Unverified

Extraction coverage

What we read of this document, and where each value was read.

Our note Jailbreak results figure-only

All 27 values were read in the document itself.

Of the 27 values, 2 are statements in words rather than numbers; they are marked * and left out of charts by default.

All 27 values were extracted for version 0 of the dataset through a web reader, which did not always reach the later sections of long PDFs (Methodology §3).