Claude Opus 4.6
35 values about Claude Opus 4.6 from 4 documents, in 9 metric families. In each family, values from its own document come first.
- Developer
- Anthropic
- Release date
- Feb 2026The date of its first card in the dataset, the Claude Opus 4.6 System Card.
- Availability
- Public
- Its own document
- Claude Opus 4.6 System Card · Feb 2026
- Also reported in
- Claude Sonnet 4.6 System Card · 17 Feb 2026
- Claude Mythos Preview System Card · 7 Apr 2026
- Claude Opus 4.7 System Card · 16 Apr 2026
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 other documents | ||
| Automated behavioral auditShare of transcripts verbalizing eval awarenessClaude Opus 4.7 System Card16 Apr 2026First reported | Claude Opus 4.7 System Card16 Apr 2026First reported | |
M2 · Reward hacking 3 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From other documents | ||
| GUI computer-use impossible tasksReward hack rateCondition: anti-hack promptClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
| Impossible tasksReward hack rateCondition: anti-hack promptClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
| Impossible tasksReward hack rateCondition: no anti-hack promptClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
M3 · Sabotage and sandbagging 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4.6 System Card · Feb 2026 | ||
| Sandbagging transcript reviewExplicit sandbagging instances foundCondition: 1,000 eval transcripts | Claude Opus 4.6 System CardFeb 2026 | |
| From other documents | ||
| SHADE-ArenaStealth success rateCondition: extended thinkingClaude Opus 4.7 System Card16 Apr 2026First reported | Claude Opus 4.7 System Card16 Apr 2026First reported | |
M4 · Misalignment audits 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From other documents | ||
| Destructive actions (synthetic backend)Destructive action rateClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
M5 · Honesty and hallucination 4 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4.6 System Card · Feb 2026 | ||
| Honesty evaluationWin rateCondition: full thinking | Claude Opus 4.6 System CardFeb 2026 | |
| From other documents | ||
| 100Q-HardCorrect answer rateClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
| False-premise honestyHonest pushback rateClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
| MASKHonesty rate under pressureClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
M6 · Sycophancy 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From other documents | ||
| Automated behavioral auditSycophancy scoreClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
M7 · Harmful compliance and over-refusal 9 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4.6 System Card · Feb 2026 | ||
| Higher-difficulty benign requestsOver-refusal rateCondition: overall across thinking modes | Claude Opus 4.6 System CardFeb 2026 | |
| Higher-difficulty violative requestsHarmless response rateCondition: overall across thinking modes | Claude Opus 4.6 System CardFeb 2026 | |
| Malicious Claude Code useRefusal rate, malicious requestsCondition: with system prompt + FileRead reminder | Claude Opus 4.6 System CardFeb 2026 | |
| Malicious computer useRefusal rate | Claude Opus 4.6 System CardFeb 2026 | |
| Single-turn benign requestsOver-refusal rateCondition: overall across thinking modesRestated later: compare with the later value | Claude Opus 4.6 System CardFeb 2026Restated later: compare with the later value | |
| Single-turn violative requestsHarmless response rateCondition: overall across thinking modesRestated later: compare with the later value | Claude Opus 4.6 System CardFeb 2026Restated later: compare with the later value | |
| From other documents | ||
| Single-turn benign requestsOver-refusal rateCondition: overall across thinking modesClaude Sonnet 4.6 System Card17 Feb 2026Restated: compare with the earlier value | Claude Sonnet 4.6 System Card17 Feb 2026Restated: compare with the earlier value | |
| Single-turn violative requestsHarmless response rateCondition: overall across thinking modesClaude Sonnet 4.6 System Card17 Feb 2026Restated: compare with the earlier value | Claude Sonnet 4.6 System Card17 Feb 2026Restated: compare with the earlier value | |
| Suicide/self-harm multi-turnAppropriate response rateClaude Sonnet 4.6 System Card17 Feb 2026First reported | Claude Sonnet 4.6 System Card17 Feb 2026First reported | |
M9 · Prompt injection 4 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4.6 System Card · Feb 2026 | ||
| Agent Red Teaming (ART)Attack success rate within k=100 attemptsCondition: extended thinkingRun by Gray Swan | Claude Opus 4.6 System CardFeb 2026 | |
| Agent Red Teaming (ART)Attack success rate within k=100 attemptsCondition: no extended thinkingRun by Gray Swan | Claude Opus 4.6 System CardFeb 2026 | |
| Browser use prompt injectionAttack success rate per attacked session | Claude Opus 4.6 System CardFeb 2026 | |
| Coding prompt injection (adaptive attacker)Attack success rateCondition: all conditions | Claude Opus 4.6 System CardFeb 2026 | |
M10 · Dangerous capabilities and risk determinations 10 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 4.6 System Card · Feb 2026 | ||
| AI R&D-4 thresholdThreshold crossed | Claude Opus 4.6 System CardFeb 2026 | |
| CBRN-4 thresholdThreshold crossed | Claude Opus 4.6 System CardFeb 2026 | |
| CybenchSuccess rate, pass@30 | Claude Opus 4.6 System CardFeb 2026 | |
| CyberGymTargeted vuln reproduction rateRestated later: compare with the later value | Claude Opus 4.6 System CardFeb 2026Restated later: compare with the later value | |
| Internal AI R&D surveyRespondents saying model could automate entry-level researcherCondition: of 16 respondents | Claude Opus 4.6 System CardFeb 2026 | |
| RSP deployment standardASL standard deployed under | Claude Opus 4.6 System CardFeb 2026 | |
| From other documents | ||
| AI R&D kernel taskBest speedupClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
| AI R&D LLM training taskAverage speedupClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
| CyberGymTargeted vuln reproduction rateClaude Mythos Preview System Card7 Apr 2026Restated: compare with the earlier value | Claude Mythos Preview System Card7 Apr 2026Restated: compare with the earlier value | |
| Multimodal virology (VCT)ScoreClaude Mythos Preview System Card7 Apr 2026First reported | Claude Mythos Preview System Card7 Apr 2026First reported | |
Restated in later documents
A later document reported values about Claude Opus 4.6 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.
| Earlier value | Later value |
|---|---|
| M7 · Harmful compliance and over-refusal | |
| Single-turn benign requestsOver-refusal rateCondition: overall across thinking modes | |
| Claude Opus 4.6 System CardFeb 2026 · its own document | Claude Sonnet 4.6 System Card17 Feb 2026No reason stated |
| Single-turn violative requestsHarmless response rateCondition: overall across thinking modes | |
| Claude Opus 4.6 System CardFeb 2026 · its own document | Claude Sonnet 4.6 System Card17 Feb 2026No reason stated |
| M10 · Dangerous capabilities and risk determinations | |
| CyberGymTargeted vuln reproduction rate | |
| Claude Opus 4.6 System CardFeb 2026 · its own document | Claude Mythos Preview System Card7 Apr 2026No reason stated |
Risk determinations
The developer's formal decisions about Claude Opus 4.6 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 4.6 System Card · Feb 2026 | Responsible Scaling Policy Claude Opus 4.6 System Card · Feb 2026 |
| Bio/chem | not crossed Responsible Scaling Policy Claude Opus 4.6 System Card · Feb 2026 | Responsible Scaling Policy Claude Opus 4.6 System Card · Feb 2026 |
| AI R&D / autonomy | not crossed Responsible Scaling Policy Claude Opus 4.6 System Card · Feb 2026 | Responsible Scaling Policy Claude Opus 4.6 System Card · Feb 2026 |