Model profile
Claude Opus 4
Evidence summary
Claude Opus 4 has an estimated overall rank of #6; its 90% source-sensitivity interval is #3–#56. Its behavior-only rank is #9; company governance moves the combined estimate to #6. Published evidence spans 19 evals and 7 of 7 behavior components. Its strongest relative result is Confabulations (confabulation_rate, #1 of 52); its weakest is Anthropic Claude Opus 4.1 System Card Addendum (harmful_request_safety, #2 of 2).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- OpenRouter ↗anthropic/claude-opus-4
- System card ↗Family-level model document · Anthropic · first party
- Release source ↗direct research preview identity date
Published eval results
Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Better | Source |
|---|---|---|---|---|---|
| AIRBench 2024 Safety Scenariossafety_scenarios | #11 / 80 | 0.857 | ↑ higher | Source ↗official | |
| Anthropic Agentic Misalignment — blackmailmisaligned_action_rate_pct | #15 / 16 | 96 | ↓ lower | Source ↗official | |
| Anthropic Agentic Misalignment — corporate espionagemisaligned_action_rate_pct | #8 / 16 | 57 | ↓ lower | Source ↗official | |
| Anthropic Agentic Misalignment — lethal actionmisaligned_action_rate_pct | #5 / 10 | 65 | ↓ lower | Source ↗official | |
| Anthropic Claude 4 System Cardagentic_coding_safety | #2 / 3 | 0.88 | ↑ higher | Source ↗official | |
| Anthropic Claude 4 System Cardbenign_request_refusal | #1 / 3 | 0.0007 | ↓ lower | Source ↗official | |
| Anthropic Claude 4 System Cardharmful_request_safety | #3 / 3 | 0.9843 | ↑ higher | Source ↗official | |
| Anthropic Claude 4 System Cardstrongreject_jailbreak_success | #2 / 3 | 0.0714 | ↓ lower | Source ↗official | |
| Anthropic Claude Opus 4.1 System Card Addendumbbq_disambiguated_accuracy | #1 / 2 | 0.911 | ↑ higher | Source ↗official | |
| Anthropic Claude Opus 4.1 System Card Addendumbenign_request_refusal | #1 / 2 | 0.0005 | ↓ lower | Source ↗official | |
| Anthropic Claude Opus 4.1 System Card Addendumharmful_request_safety | #2 / 2 | 0.9727 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #43 / 105 | 0.34 | ↑ higher | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #16 / 105 | 4.736 | ↓ lower | Source ↗official | |
| Confabulationsconfabulation_rate | #1 / 52 | 2.723 | ↓ lower | Source ↗official | |
| Emergent Collusionhigh_illegality_game_rate | #6 / 13 | 0.36 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #27 / 260 | 43.41 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #236 / 260 | 75 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #17 / 260 | 98.89 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #41 / 258 | 99.18 | ↑ higher | Source ↗official | |
| FORTRESSaverage_risk_score | #22 / 49 | 26.19 | ↓ lower | Source ↗official | |
| FORTRESSover_refusal_score | #13 / 46 | 2.205 | ↓ lower | Source ↗official | |
| HELM Safetyanthropic_red_team | #40 / 80 | 0.989 | ↑ higher | Source ↗official | |
| HELM Safetybbq | #4 / 80 | 0.982 | ↑ higher | Source ↗official | |
| HELM Safetyharmbench | #21 / 80 | 0.904 | ↑ higher | Source ↗official | |
| HELM Safetysimple_safety_tests | #26 / 80 | 0.9975 | ↑ higher | Source ↗official | |
| HELM Safetyxstest | #24 / 80 | 0.97 | ↑ higher | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #42 / 54 | 24.9 | ↑ higher | Source ↗official | |
| MASKlying_probability_pct | #12 / 53 | 15.93 | ↓ lower | Source ↗official | |
| Social Welfare Function Benchmarkfairness | #5 / 19 | 0.547 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #71 / 102 | 6.58 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #37 / 102 | 4.42 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #22 / 102 | 4.53 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #49 / 102 | 1.98 | ↓ lower | Source ↗official |
Values evaluations
Descriptive values and political-framing results are separate from safety/ethics ranks. Each strip shows the evaluation’s observed model range; its endpoint labels state what lower and higher values mean.