Model profile
Claude Sonnet 4.6
Anthropicdeveloper
2026-01-21release date
#14 / 267overall rank
25eval lineages
1discovery sources
Evidence summary
Claude Sonnet 4.6 has an estimated overall rank of #14; its 90% source-sensitivity interval is #4–#72. Its behavior-only rank is #17; company governance moves the combined estimate to #14. Published evidence spans 25 evals and 7 of 7 behavior components. Its strongest relative result is SM-Bench (anti_hallucination, #1 of 73); its weakest is Claude Sonnet 4.6 User Wellbeing (child_multiturn_appropriate_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-sonnet-4-6
- OpenRouter ↗anthropic/claude-sonnet-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 | #41 / 311 | 0.4614 | ↓ lower | Source ↗official | |
| AgentAbstainabstain | #4 / 17 | 66.4 | ↑ higher | Source ↗official | |
| AgentAbstaincar | #4 / 17 | 65.4 | ↑ higher | Source ↗official | |
| AgentAbstainpaired | #3 / 17 | 53.4 | ↑ higher | Source ↗official | |
| ANIMAscore | #17 / 19 | 0.5646 | ↑ higher | Source ↗official | |
| BioSecBench-Refusalbalanced_refusal_score | #2 / 10 | 0.4304 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #2 / 105 | 0.9 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #12 / 43 | 54.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #6 / 48 | 31.5 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #10 / 49 | 43.1 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #17 / 45 | 84.9 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #14 / 51 | 11.7 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #17 / 48 | 16.5 | ↓ lower | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #9 / 105 | 3.041 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 Overrefusalhigher_difficulty_overrefusal_rate | #2 / 5 | 0.18 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 Overrefusaloverall_overrefusal_rate | #2 / 5 | 0.23 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingchild_benign_refusal_rate | #2 / 4 | 0.08 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingchild_multiturn_appropriate_rate | #4 / 4 | 95 | ↑ higher | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingchild_violative_harmless_rate | #1 / 4 | 99.96 | ↑ higher | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingselfharm_benign_refusal_rate | #3 / 4 | 0.17 | ↓ lower | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingselfharm_harmless_rate | #2 / 4 | 99.73 | ↑ higher | Source ↗official | |
| Claude Sonnet 4.6 User Wellbeingselfharm_multiturn_appropriate_rate | #1 / 4 | 98 | ↑ higher | Source ↗official | |
| DystopiaBenchbasaglia_score | #2 / 50 | 24.2 | ↓ lower | Source ↗official | |
| DystopiaBenchbaudrillard_score | #6 / 50 | 27.87 | ↓ lower | Source ↗official | |
| DystopiaBenchhuxley_score | #3 / 50 | 17.6 | ↓ lower | Source ↗official | |
| DystopiaBenchlaguardia_score | #3 / 50 | 33 | ↓ lower | Source ↗official | |
| DystopiaBenchorwell_score | #5 / 50 | 24.57 | ↓ lower | Source ↗official | |
| DystopiaBenchpetrov_score | #3 / 50 | 24.93 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #101 / 260 | 18.86 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #184 / 260 | 84.67 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #7 / 260 | 99.44 | ↑ higher | Source ↗official | |
| LiveSecBenchethics | #8 / 43 | 78.63 | ↑ higher | Source ↗official | |
| LiveSecBenchfactuality | #1 / 43 | 92.17 | ↑ higher | Source ↗official | |
| LiveSecBenchlegality | #2 / 43 | 97.77 | ↑ higher | Source ↗official | |
| LiveSecBenchprivacy | #3 / 43 | 89.26 | ↑ higher | Source ↗official | |
| LiveSecBenchpsychological_health | #8 / 43 | 72.01 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #17 / 50 | 84.9 | ↓ lower | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #3 / 31 | 4.2 | ↓ lower | Source ↗official | |
| Manager Coercion Benchfabrication_rate | #1 / 13 | 0 | ↓ lower | Source ↗official | |
| MORUscore | #11 / 13 | 69.73 | ↑ higher | Source ↗official | |
| PHAREbias_resistance_diagnostic | #42 / 66 | 0.4128 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #8 / 70 | 0.8551 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #2 / 70 | 0.9993 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #13 / 67 | 0.6939 | ↑ higher | Source ↗official | |
| RefusalBenchyouden_j | #5 / 19 | 0.6766 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #45 / 73 | 80 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #19 / 73 | 88.99 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #1 / 73 | 100 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #73 / 73 | 33.43 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #10 / 73 | 91.53 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #53 / 102 | 6.78 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #63 / 102 | 4.65 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #38 / 102 | 4.67 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #57 / 102 | 2.1 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #9 / 68 | 37.82 | ↑ higher | Source ↗official | |
| VETO Misfired Alignmentmisfired_alignment_rate_pct | #21 / 25 | 10.9 | ↓ lower | Source ↗official | |
| Vigil Mental Health Safetyoverall_score | #1 / 23 | 83 | ↑ higher | 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.
UGI Political Values
Agent-ValueBench Moral Foundations (MFT08)
Agent-ValueBench HEXACO
| Dimension | Value | Distribution |
|---|---|---|
| Openness to experience | 5.8 | |
| Honesty-humility | 7.3 | |
| Extraversion | 5.4 | |
| Agreeableness | 6.1 | |
| Conscientiousness | 7.1 |