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.

M5 · Honesty and hallucination 2 values

M5 Honesty and hallucination: values about Claude Opus 4.5
Evaluation and metricValueDocument
From its own documentClaude Opus 4.5 System Card · 24 Nov 2025
AA-OmniscienceImprovement from 16k thinking vs no thinkingCondition: 16k thinking budget; extended thinkingClaude Opus 4.5 System Card24 Nov 2025
SimpleQA VerifiedImprovement in correct answers vs prior ClaudeCondition: 16k thinking budget; extended thinkingClaude Opus 4.5 System Card24 Nov 2025

M6 · Sycophancy 1 value

M6 Sycophancy: values about Claude Opus 4.5
Evaluation and metricValueDocument
From its own documentClaude Opus 4.5 System Card · 24 Nov 2025
Sycophancy in prefilled conversationsCourse-correction rateCondition: continuing sycophantic conversationClaude Opus 4.5 System Card24 Nov 2025

M7 · Harmful compliance and over-refusal 13 values

M7 Harmful compliance and over-refusal: values about Claude Opus 4.5
Evaluation and metricValueDocument
From its own documentClaude Opus 4.5 System Card · 24 Nov 2025
Malicious agentic codingSafety scoreCondition: without safeguardsClaude Opus 4.5 System Card24 Nov 2025
Malicious Claude Code useRefusal rate, malicious requestsCondition: with mitigationsClaude Opus 4.5 System Card24 Nov 2025
Malicious Claude Code useRefusal rate, malicious requestsCondition: without safeguardsClaude Opus 4.5 System Card24 Nov 2025
Malicious Claude Code useSuccess rate, dual-use & benignCondition: without safeguardsClaude Opus 4.5 System Card24 Nov 2025
Malicious computer useRefusal rateCondition: without safeguardsClaude Opus 4.5 System Card24 Nov 2025
Multi-turn testingRubric failure rate, deadly weaponsClaude Opus 4.5 System Card24 Nov 2025
Multi-turn testingRubric failure rate, violent extremismClaude Opus 4.5 System Card24 Nov 2025
Single-turn benign requestsOver-refusal rateCondition: overall across thinking modes; all languagesRestated later: compare with the later valueClaude Opus 4.5 System Card24 Nov 2025Restated later: compare with the later value
Single-turn violative requestsHarmless response rateCondition: extended thinking; all languagesClaude Opus 4.5 System Card24 Nov 2025
Single-turn violative requestsHarmless response rateCondition: no extended thinking; all languagesClaude Opus 4.5 System Card24 Nov 2025
Single-turn violative requestsHarmless response rateCondition: overall across thinking modes; all languagesRestated later: compare with the later valueClaude Opus 4.5 System Card24 Nov 2025Restated later: compare with the later value
From other documents
Single-turn benign requestsOver-refusal rateCondition: overall across thinking modesClaude Opus 4.6 System CardFeb 2026Restated: compare with the earlier valueClaude Opus 4.6 System CardFeb 2026Restated: compare with the earlier value
Single-turn violative requestsHarmless response rateCondition: overall across thinking modesClaude Opus 4.6 System CardFeb 2026Restated: compare with the earlier valueClaude Opus 4.6 System CardFeb 2026Restated: compare with the earlier value

M9 · Prompt injection 10 values

M9 Prompt injection: values about Claude Opus 4.5
Evaluation and metricValueDocument
From its own documentClaude Opus 4.5 System Card · 24 Nov 2025
Agent Red Teaming (ART)Attack success rateCondition: k=1Run by Gray SwanClaude Opus 4.5 System Card24 Nov 2025
Coding prompt injection (adaptive attacker)Attack success rateCondition: extended thinking; without safeguards; 1 attemptClaude Opus 4.5 System Card24 Nov 2025
Coding prompt injection (adaptive attacker)Attack success rateCondition: extended thinking; without safeguards; 200 attemptsClaude Opus 4.5 System Card24 Nov 2025
Coding prompt injection (adaptive attacker)Attack success rateCondition: no extended thinking; without safeguards; 1 attemptClaude Opus 4.5 System Card24 Nov 2025
Coding prompt injection (adaptive attacker)Attack success rateCondition: no extended thinking; without safeguards; 200 attemptsClaude Opus 4.5 System Card24 Nov 2025
Computer use prompt injection (adaptive attacker)Attack success rateCondition: extended thinking; without safeguards; 1 attemptClaude Opus 4.5 System Card24 Nov 2025
Computer use prompt injection (adaptive attacker)Attack success rateCondition: extended thinking; without safeguards; 200 attemptsClaude Opus 4.5 System Card24 Nov 2025
Computer use prompt injection (adaptive attacker)Attack success rateCondition: no extended thinking; with safeguards; 200 attemptsClaude Opus 4.5 System Card24 Nov 2025
Computer use prompt injection (adaptive attacker)Attack success rateCondition: no extended thinking; without safeguards; 1 attemptClaude Opus 4.5 System Card24 Nov 2025
Computer use prompt injection (adaptive attacker)Attack success rateCondition: no extended thinking; without safeguards; 200 attemptsClaude Opus 4.5 System Card24 Nov 2025

M10 · Dangerous capabilities and risk determinations 8 values

M10 Dangerous capabilities and risk determinations: values about Claude Opus 4.5
Evaluation and metricValueDocument
From its own documentClaude Opus 4.5 System Card · 24 Nov 2025
AI R&D-4 thresholdThreshold crossedClaude Opus 4.5 System Card24 Nov 2025
ASL-4 virology uplift trialProtocol-score uplift vs internet-onlyClaude Opus 4.5 System Card24 Nov 2025
Bioinformatics evaluationsScoreClaude Opus 4.5 System Card24 Nov 2025
CBRN-4 thresholdThreshold crossedClaude Opus 4.5 System Card24 Nov 2025
Internal AI R&D suite 2Suite scoreClaude Opus 4.5 System Card24 Nov 2025
Internal model use surveyStaff reporting >=100% productivity gainCondition: of 18 staff surveyedClaude Opus 4.5 System Card24 Nov 2025
RSP deployment standardASL standard deployed underClaude Opus 4.5 System Card24 Nov 2025
SWE-bench Verified (hard subset)Problems solvedCondition: of 45 problemsClaude Opus 4.5 System Card24 Nov 2025

Restated in later documents

A later document reported values about Claude Opus 4.5 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 Claude Opus 4.5 restated in later documents
Earlier valueLater value
M7 · Harmful compliance and over-refusal
Single-turn benign requestsOver-refusal rateCondition: overall across thinking modes; all languagesLater condition: overall across thinking modes
Claude Opus 4.5 System Card24 Nov 2025 · its own documentClaude Opus 4.6 System CardFeb 2026No reason stated
Single-turn violative requestsHarmless response rateCondition: overall across thinking modes; all languagesLater condition: overall across thinking modes
Claude Opus 4.5 System Card24 Nov 2025 · its own documentClaude Opus 4.6 System CardFeb 2026No reason stated

Risk determinations

The developer's formal decisions about Claude Opus 4.5 under its framework, as printed. Levels from different frameworks do not map onto one another.

Risk determinations about Claude Opus 4.5
DomainLevel as printedFramework and document
OverallASL-3 Responsible Scaling Policy Claude Opus 4.5 System Card · 24 Nov 2025Responsible Scaling Policy Claude Opus 4.5 System Card · 24 Nov 2025
Bio/chemNo Responsible Scaling Policy Claude Opus 4.5 System Card · 24 Nov 2025Responsible Scaling Policy Claude Opus 4.5 System Card · 24 Nov 2025
AI R&D / autonomyNo Responsible Scaling Policy Claude Opus 4.5 System Card · 24 Nov 2025Responsible Scaling Policy Claude Opus 4.5 System Card · 24 Nov 2025