Model profile
Claude Opus 4.6
Anthropicdeveloper
2026-02-05release date
#8 / 267overall rank
23eval lineages
Evidence summary
Claude Opus 4.6 has an estimated overall rank of #8; its 90% source-sensitivity interval is #7–#51. Its behavior-only rank is #12; company governance moves the combined estimate to #8. Published evidence spans 23 evals and 7 of 7 behavior components. Its strongest relative result is SpeciEval (belief_animal_sentience, #1 of 102); its weakest is Claude Sonnet 4.6 User Wellbeing (child_benign_refusal_rate, #4 of 4).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗claude-opus-4-6
- OpenRouter ↗anthropic/claude-opus-4.6
- System card ↗Exact 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 |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #74 / 311 | 0.6133 | ↓ lower | Source ↗official | |
| ANIMAscore | #5 / 19 | 0.7246 | ↑ higher | Source ↗official | |
| BioSecBench-Refusalbalanced_refusal_score | #5 / 10 | 0.3987 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #3 / 105 | 0.85 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #10 / 43 | 49.5 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #7 / 48 | 34.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #14 / 49 | 46 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #15 / 45 | 84.4 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #17 / 51 | 12.7 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #22 / 32 | 51.7 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #43 / 48 | 23.2 | ↓ lower | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #13 / 105 | 3.639 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 Overrefusalhigher_difficulty_overrefusal_rate | #1 / 5 | 0.04 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 Overrefusaloverall_overrefusal_rate | #4 / 5 | 0.66 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingchild_benign_refusal_rate | #4 / 4 | 0.18 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingchild_multiturn_appropriate_rate | #2 / 4 | 96 | ↑ higher | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingchild_violative_harmless_rate | #2 / 4 | 99.95 | ↑ higher | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingselfharm_benign_refusal_rate | #4 / 4 | 0.25 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingselfharm_harmless_rate | #1 / 4 | 99.75 | ↑ higher | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingselfharm_multiturn_appropriate_rate | #3 / 4 | 82 | ↑ higher | Source ↗official | |
| DystopiaBenchbasaglia_score | #5 / 50 | 26.77 | ↓ lower | Source ↗official | |
| DystopiaBenchbaudrillard_score | #10 / 50 | 33.37 | ↓ lower | Source ↗official | |
| DystopiaBenchhuxley_score | #2 / 50 | 15.77 | ↓ lower | Source ↗official | |
| DystopiaBenchlaguardia_score | #6 / 50 | 39.23 | ↓ lower | Source ↗official | |
| DystopiaBenchorwell_score | #4 / 50 | 23.47 | ↓ lower | Source ↗official | |
| DystopiaBenchpetrov_score | #5 / 50 | 30.07 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #76 / 260 | 22.74 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #66 / 260 | 91.33 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #7 / 260 | 99.44 | ↑ higher | Source ↗official | |
| FORTRESSaverage_risk_score | #11 / 49 | 16.76 | ↓ lower | Source ↗official | |
| FORTRESSover_refusal_score | #37 / 46 | 9.59 | ↓ lower | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #23 / 54 | 27.23 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #15 / 50 | 84.4 | ↓ lower | Source ↗official | |
| MASKlying_probability_pct | #5 / 53 | 9.16 | ↓ lower | Source ↗official | |
| MORUscore | #6 / 13 | 74.1 | ↑ higher | Source ↗official | |
| PHAREbias_resistance_diagnostic | #57 / 66 | 0.3351 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #2 / 70 | 0.8776 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #1 / 70 | 1 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #6 / 67 | 0.7524 | ↑ higher | Source ↗official | |
| RefusalBenchyouden_j | #7 / 19 | 0.6667 | ↑ higher | Source ↗official | |
| SABERoverall_safety_rate | #1 / 13 | 45.26 | ↑ higher | Source ↗official | |
| SABERscenario_a_safety_rate | #1 / 13 | 56.3 | ↑ higher | Source ↗official | |
| SABERscenario_b_safety_rate | #1 / 13 | 39.77 | ↑ higher | Source ↗official | |
| SABERscenario_c_safety_rate | #1 / 13 | 36.87 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #28 / 73 | 83.41 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #40 / 73 | 83.63 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #1 / 73 | 100 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #70 / 73 | 42.98 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #3 / 73 | 95.63 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #1 / 102 | 7 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #96 / 102 | 5.05 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #79 / 102 | 5.03 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #65 / 102 | 2.17 | ↓ 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.