Claude Opus 4
36 values about Claude Opus 4 from 3 documents, in 10 metric families. In each family, values from its own document come first.
- Developer
- Anthropic
- Release date
- 22 May 2025The date of its first card in the dataset, the Claude Opus 4 & Sonnet 4 System Card.
- Availability
- Public
- Its own document
- Claude Opus 4 & Sonnet 4 System Card · 22 May 2025
- Also reported in
- Claude Opus 4.1 System Card (addendum) · 5 Aug 2025
- Claude Sonnet 4.5 System Card · 29 Sep 2025
Data as of 26 Sep 2026 · Dataset v0.1 · Methodology v0.1 · Changelog
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.
M1 · Evaluation awareness 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Automated behavioral auditTranscripts flagged for situational awarenessCondition: of 414 audit transcripts | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
M2 · Reward hacking 8 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Reward hacking (hard-coding)Avg decrease in hard-coding vs Sonnet 3.7Condition: vs Sonnet 3.7 | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| From other documents | ||
| Impossible tasksClassifier hack rateCondition: anti-hack promptClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| Impossible tasksClassifier hack rateCondition: no anti-hack promptClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| Reward-hack-prone coding tasksClassifier hack rateCondition: no anti-hack promptClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| Reward-hack-prone coding tasksHidden-test hack rateCondition: no anti-hack promptClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| Impossible tasksClassifier hack rateCondition: anti-hack promptClaude Sonnet 4.5 System Card29 Sep 2025First reported | Claude Sonnet 4.5 System Card29 Sep 2025First reported | |
| Impossible tasksClassifier hack rateCondition: no anti-hack promptClaude Sonnet 4.5 System Card29 Sep 2025First reported | Claude Sonnet 4.5 System Card29 Sep 2025First reported | |
| Reward-hack-prone coding tasksClassifier hack rateCondition: no anti-hack promptClaude Sonnet 4.5 System Card29 Sep 2025First reported | Claude Sonnet 4.5 System Card29 Sep 2025First reported | |
M3 · Sabotage and sandbagging 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Subtle sabotage (agentic)Side-task success without monitor detectionCondition: hidden scratchpad | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Subtle sabotage (agentic)Side-task success without monitor detectionCondition: scratchpad visible to monitor | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
M4 · Misalignment audits 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Automated behavioral auditSurprising high-impact actions foundCondition: audit transcripts | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Opportunistic blackmail scenarioShare of rollouts with blackmailCondition: replacement model shares values | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
M5 · Honesty and hallucination 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Reasoning-behavior monitorKnowingly hallucinated informationCondition: RL-style transcripts | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Reasoning-behavior monitorOutputs showing deception (overall)Condition: RL-style transcripts | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
M6 · Sycophancy 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Reasoning-behavior monitorOutputs showing sycophancyCondition: RL-style transcripts | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
M7 · Harmful compliance and over-refusal 10 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Malicious agentic codingSafety scoreCondition: without safeguards | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Single-turn benign requestsOver-refusal rateCondition: overall across thinking modes | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Single-turn violative requestsHarmless response rateCondition: extended thinking | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Single-turn violative requestsHarmless response rateCondition: no extended thinking | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Single-turn violative requestsHarmless response rateCondition: overall across thinking modes | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Single-turn violative requestsHarmless response rateCondition: overall across thinking modes; with ASL-3 safeguards | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| From other documents | ||
| Single-turn benign requestsOver-refusal rateCondition: overall across thinking modesClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| Single-turn violative requestsHarmless response rateCondition: overall across thinking modesClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| Single-turn benign requestsOver-refusal rateCondition: overall across thinking modesClaude Sonnet 4.5 System Card29 Sep 2025Restated | Claude Sonnet 4.5 System Card29 Sep 2025Restated | |
| Single-turn violative requestsHarmless response rateCondition: overall across thinking modesClaude Sonnet 4.5 System Card29 Sep 2025Restated | Claude Sonnet 4.5 System Card29 Sep 2025Restated | |
M8 · Jailbreak robustness 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| StrongREJECTBest-jailbreak StrongREJECT scoreCondition: extended thinking | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| StrongREJECTBest-jailbreak StrongREJECT scoreCondition: no extended thinking | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
M9 · Prompt injection 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| Computer-use prompt injectionAttack prevention rateCondition: with safeguards | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| Computer-use prompt injectionAttack prevention rateCondition: without safeguards | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
M10 · Dangerous capabilities and risk determinations 6 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4 & Sonnet 4 System Card · 22 May 2025 | ||
| CybenchCTF challenges solvedCondition: of 39 challenges | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| RSP deployment standardASL standard deployed under | Claude Opus 4 & Sonnet 4 System Card22 May 2025 | |
| From other documents | ||
| AI R&D kernel taskBest speedupClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| CybenchCTF challenges solvedCondition: of 35-challenge subsetClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| Long-form virology task 1ScoreClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
| SWE-bench Verified (hard subset)Problems solved, pass@1 avgCondition: of 42 problemsClaude Opus 4.1 System Card (addendum)5 Aug 2025First reported | Claude Opus 4.1 System Card (addendum)5 Aug 2025First reported | |
Risk determinations
The developer's formal decisions about Claude Opus 4 under its framework, as printed. Levels from different frameworks do not map onto one another.
| Domain | Level as printed | Framework and document |
|---|---|---|
| Overall | ASL-3 Responsible Scaling Policy Claude Opus 4 & Sonnet 4 System Card · 22 May 2025 | Responsible Scaling Policy Claude Opus 4 & Sonnet 4 System Card · 22 May 2025 |