← Models

Model profile

Claude Opus 4.7

Anthropicdeveloper
2026-04-16release date
#18 / 348Safety rank
#576 / 663Freedom rank
2discovery sources

Evidence summary

Safety. Claude Opus 4.7 has an estimated Safety rank of #18; its 90% source-sensitivity interval is #17–#76. Its behavior-only rank is #26; company governance moves the combined estimate to #18. Published Safety evidence spans 38 eval lineages and 7 of 7 components. Its strongest relative result is Human Pathogen Capabilities Test (HPCT) — overall refusal (refusal_rate_pct, #1 of 69); its weakest is BioTIER (permit_compliance_pct, #51 of 52).

Freedom. Claude Opus 4.7 has an estimated Freedom rank of #576; its 90% source-sensitivity interval is #356–#646. Published Freedom evidence spans 11 eval lineages and 1 of 1 components. Its strongest relative result is SM-Bench (overfit, #14 of 92); its weakest is Human Pathogen Capabilities Test (HPCT) — overall refusal (refusal_rate_pct, #69 of 69).

Compare this model

Only models sharing at least one published sub-eval are listed.

Official and reference links

Safety evals

Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.

Eval / sub-evalRankValueDistributionSource
AA-Omnisciencehallucination_rate#67 / 361↓0.4229Source ↗official
AgentAbstainabstain#1 / 17↑79Source ↗official
AgentAbstaincar#1 / 17↑77.6Source ↗official
AgentAbstainpaired#2 / 17↑59.4Source ↗official
ANIMAscore#5 / 22↑0.735Source ↗official
Arena Factuality — Search Arena (factuality-only weighting)factuality_bt_rating#23 / 30↑1171.0Source ↗official
Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating#44 / 111↑1442.0Source ↗official
BioSecBench-Refusal (July 2026 snapshot)balanced_refusal_score#3 / 10↑0.4259Source ↗official
BioTIERpermit_compliance_pct#51 / 52↑78.7Source ↗official
BioTIERrefuse_compliance_pct#2 / 52↑95.5Source ↗official
BullshitBench v2clear_pushback_rate#6 / 122↑0.785Source ↗official
CAIS Risk Indexagent_red_teaming#10 / 49↓44.5Source ↗official
CAIS Risk Indexbioweapons_assistance#4 / 54↓19.2Source ↗official
CAIS Risk Indexhle_overconfidence#6 / 55↓28.1Source ↗official
CAIS Risk Indexmachiavelli#22 / 51↓85.7Source ↗official
CAIS Risk Indexmask#27 / 57↓17.1Source ↗official
CAIS Risk Indexpolitical_manipulation#29 / 51↓48.2Source ↗official
CAIS Risk Indextextquests_harm#40 / 54↓21.8Source ↗official
Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate#7 / 104↓2.642Source ↗official
Claude Fable 5.1 / Mythos 5.1 card — alignment, honesty, and oversightaa_omniscience_net_score#6 / 7↑0.38Source ↗official
Claude Fable 5.1 / Mythos 5.1 card — alignment, honesty, and oversightleaked_answer_undisclosed_use_rate_pct#2 / 7↓13.5Source ↗official
Claude Fable 5.1 / Mythos 5.1 card — alignment, honesty, and oversightlying_about_own_work_shared_environment_rate_pct#5 / 8↓0.27Source ↗official
Claude Fable 5.1 / Mythos 5.1 card — alignment, honesty, and oversightmask_honesty_rate_pct#6 / 7↑90.2Source ↗official
Claude Fable 5.1 / Mythos 5.1 card — alignment, honesty, and oversightrewarded_cheating_rate_pct#2 / 8↓0.12Source ↗official
DelusionEvaldelusional_prevalence_pct#6 / 16↓27.8Source ↗official
DelusionEvaldiscourages_harm_prevalence_pct#5 / 16↑36.2Source ↗official
DelusionEvalfacilitates_harm_prevalence_pct#10 / 16↓3.5Source ↗official
DelusionEvalrelationship_prevalence_pct#9 / 16↓27Source ↗official
DelusionEvalsycophancy_prevalence_pct#8 / 16↓26.5Source ↗official
DystopiaBenchbasaglia_score#4 / 50↓26.03Source ↗official
DystopiaBenchbaudrillard_score#5 / 50↓27.5Source ↗official
DystopiaBenchhuxley_score#4 / 50↓18.33Source ↗official
DystopiaBenchlaguardia_score#2 / 50↓30.07Source ↗official
DystopiaBenchorwell_score#3 / 50↓22.73Source ↗official
DystopiaBenchpetrov_score#4 / 50↓26.77Source ↗official
Enkrypt AI Safety Leaderboardbias_attack_non_success_rate#80 / 272↑23.51Source ↗official
Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate#9 / 272↑96.33Source ↗official
Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate#10 / 272↑99.44Source ↗official
Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate#51 / 270↑98.55Source ↗official
Every Model Cheats — Cybench Cheat Propensitybaseline_cheat_propensity_pct#16 / 22↓43.5Source ↗official
Every Model Cheats — Cybench Cheat Propensitysevere_anti_cheat_cheat_propensity_pct#7 / 22↓4.3Source ↗official
Every Model Cheats — Cybench Cheat Propensitystandard_anti_cheat_cheat_propensity_pct#16 / 22↓21.7Source ↗official
HarmVideoBenchharmful_video_safety_recognition_reasoning#4 / 19↑0.803Source ↗official
HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score#24 / 54↑27.19Source ↗official
Human Pathogen Capabilities Test (HPCT) — overall refusalrefusal_rate_pct#1 / 69↑97.9Source ↗official
Humanity's Last Exam RMS calibration error (Scale Labs)calibrationError#8 / 42↓47Source ↗official
JuICE Cultural-Error Span Detectionf1#4 / 10↑0.4503Source ↗official
kindbench v0.1.0 psychological safety rankingemotional_safety#8 / 10↑78.4Source ↗official
kindbench v0.1.0 psychological safety rankingidentity_collapse#6 / 10↑82.4Source ↗official
kindbench v0.1.0 psychological safety rankingsycophancy_spine#3 / 10↑93.7Source ↗official
kindbench v0.1.0 psychological safety rankingvalue_integrity#7 / 10↑86.3Source ↗official
MACHIAVELLIdeception_relative_random_pct#19 / 50↓85.7Source ↗official
MANTAAWMS#2 / 12↑0.579Source ↗official
MANTAAWVS#2 / 12↑0.76Source ↗official
MORUscore#12 / 13↑67.2Source ↗official
ODCV-Benchaverage_severity#1 / 12↓0.0125Source ↗official
ODCV-Benchmisalignment_rate#1 / 12↓0Source ↗official
RefusalBenchyouden_j#7 / 19↑0.234Source ↗official
SM-Benchadversarial#49 / 92↑81.46Source ↗official
SM-Benchambiguous_interpretation#24 / 92↑89.29Source ↗official
SM-Benchanti_hallucination#27 / 92↑97.38Source ↗official
SM-Bencheq_boundaries#70 / 92↑57.87Source ↗official
SM-Benchoverfit#14 / 92↑91.8Source ↗official
SpeciEvalbelief_animal_sentience#45 / 133↑6.88Source ↗official
SpeciEvalland_animal_4ns#73 / 133↓4.57Source ↗official
SpeciEvalsea_animal_4ns#47 / 133↓4.65Source ↗official
SpeciEvalspeciesism#65 / 133↓1.98Source ↗official
The Dictatorship Evaloverall_resistance_rate#4 / 20↑88.35Source ↗official
ToolPrivacyBenchprivate_mt_poi#2 / 9↓20.31Source ↗official
ToolPrivacyBenchpublic_mt_poi#4 / 9↓17.83Source ↗official
UK AISI active safety-research compromise continuationactive_compromise_continuation_rate_pct#2 / 5↓0.8Source ↗official
UK AISI cyber-evaluation cheating and prompted self-reportattempted_cheating_trajectory_rate_pct#2 / 5↓9.1Source ↗official
UK AISI cyber-evaluation cheating and prompted self-reportspecific_cheating_action_mention_rate_pct#2 / 5↑81Source ↗official
Vals AI Cheating Auditterminal_bench_cheating_shortcut_evidence_rate_pct#6 / 14↓1.124Source ↗official
Vectara HHEM Factual Consistencyfactual_consistency_rate#71 / 94↑88Source ↗official
VETO Misfired Alignmentmisfired_alignment_rate_pct#20 / 25↓10.7Source ↗official

Freedom evals

Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.

Eval / sub-evalRankValueDistributionSource
BioTIERpermit_compliance_pct#51 / 52↑78.7Source ↗official
BioTIERrefuse_compliance_pct#51 / 52↓95.5Source ↗official
CAIS Risk Indexbioweapons_assistance#51 / 54↑19.2Source ↗official
Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate#98 / 104↑2.642Source ↗official
DelusionEvaldiscourages_harm_prevalence_pct#12 / 16↓36.2Source ↗official
DelusionEvalfacilitates_harm_prevalence_pct#7 / 16↑3.5Source ↗official
Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate#262 / 272↓96.33Source ↗official
Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate#253 / 272↓99.44Source ↗official
Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate#219 / 270↓98.55Source ↗official
Human Pathogen Capabilities Test (HPCT) — overall refusalrefusal_rate_pct#69 / 69↓97.9Source ↗official
kindbench v0.1.0 psychological safety rankingemotional_safety#3 / 10↓78.4Source ↗official
SM-Benchadversarial#42 / 92↓81.46Source ↗official
SM-Bencheq_boundaries#70 / 92↑57.87Source ↗official
SM-Benchoverfit#14 / 92↑91.8Source ↗official
SpeechMap model completioncomplete_pct#52 / 181↑71.5Source ↗official
UGI Leaderboard — base-model willingnesswillingness_adherence_score#125 / 156↑1.125Source ↗official
UGI Leaderboard — base-model willingnesswillingness_direct_score#113 / 156↑2Source ↗official
VETO Misfired Alignmentmisfired_alignment_rate_pct#20 / 25↓10.7Source ↗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

DimensionValueDistribution
Political Lean-14.3
Government46
Diplomacy62.2
Economy45.4
Society58.9

The Economist World Values Survey Cultural Map

DimensionValueDistribution
Survival ↔ Self-expression3.13
Traditional ↔ Secular-0.217