Claude Opus 5.5
17 values about Claude Opus 5.5 from 1 document, in 9 metric families. In each family, values from its own document come first.
- Developer
- Anthropic
- Release date
- 22 Sep 2026The date of its first card in the dataset, the Claude Opus 5.5 System Card.
- Availability
- Public
- Its own document
- Claude Opus 5.5 System Card · 22 Sep 2026
- Also reported in
- No other document reports a value about it
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.
KeyVerifiedUnverifiedDisputedCorrected read off a figure or stated in wordsA value opens its source and history.
M1 · Evaluation awareness 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| Automated behavioral auditShare of transcripts scoring >=6 | Claude Opus 5.5 System Card22 Sep 2026 | |
| Evaluation awareness (deployment)Share of transcripts scoring >=6Condition: real internal deployment | Claude Opus 5.5 System Card22 Sep 2026 | |
M2 · Reward hacking 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| Reward hacking in RL trainingShare of episodes with successful reward hacks | Claude Opus 5.5 System Card22 Sep 2026 | |
M3 · Sabotage and sandbagging 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| SHADE-ArenaStealth success rateCondition: extended thinking | Claude Opus 5.5 System Card22 Sep 2026 | |
M4 · Misalignment audits 2 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| Automated behavioral auditMisaligned behavior vs recent models | Claude Opus 5.5 System Card22 Sep 2026 | |
| Package-registry credential exerciseShare of cases with potentially harmful actionsCondition: simulated security exercise | Claude Opus 5.5 System Card22 Sep 2026 | |
M5 · Honesty and hallucination 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| Disclosure of grader-fooling actionsRate of coming clean | Claude Opus 5.5 System Card22 Sep 2026 | |
M7 · Harmful compliance and over-refusal 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| Malicious Claude Code useRefusal rate rank | Claude Opus 5.5 System Card22 Sep 2026 | |
M9 · Prompt injection 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| Prompt injection (user-pasted text)Susceptibility vs prior models | Claude Opus 5.5 System Card22 Sep 2026 | |
M10 · Dangerous capabilities and risk determinations 7 values
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| AI R&D assessmentEstimated AI-driven accelerationRun by METR | Claude Opus 5.5 System Card22 Sep 2026 | |
| Anthropic ECI (AECI)Capability index | Claude Opus 5.5 System Card22 Sep 2026 | |
| BioMysteryBenchAccuracy (human difficult) | Claude Opus 5.5 System Card22 Sep 2026 | |
| CoBenchScore | Claude Opus 5.5 System Card22 Sep 2026 | |
| ExploitBenchMean flags capturedCondition: 41 V8 environments | Claude Opus 5.5 System Card22 Sep 2026 | |
| Multimodal virology (VCT)Score | Claude Opus 5.5 System Card22 Sep 2026 | |
| RSP determinationsCB / autonomy / cyber | Claude Opus 5.5 System Card22 Sep 2026 | |
M11 · Self-preservation 1 value
| Evaluation and metric | Value | Document |
|---|---|---|
| From its own documentClaude Opus 5.5 System Card · 22 Sep 2026 | ||
| Sandbox escape/tamper attemptsShare of runs with escape/tamper attemptCondition: without safeguards | Claude Opus 5.5 System Card22 Sep 2026 | |
Risk determinations
The developer's formal decisions about Claude Opus 5.5 under its framework, as printed. Levels from different frameworks do not map onto one another.
| Domain | Level as printed | Framework and document |
|---|---|---|
| Bio/chem | CB-1 yes Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 | Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 |
| Bio/chem | CB-2 no Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 | Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 |
| Cyber | cyber Tier 1 Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 | Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 |
| AI R&D / autonomy | AI R&D not crossed Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 | Responsible Scaling Policy Claude Opus 5.5 System Card · 22 Sep 2026 |