Model profile
Claude Opus 4.7
Anthropicdeveloper
2026-04-16release date
#5 / 267overall rank
25eval lineages
2discovery sources
Evidence summary
Claude Opus 4.7 has an estimated overall rank of #5; its 90% source-sensitivity interval is #3–#50. Its behavior-only rank is #6; company governance moves the combined estimate to #5. Published evidence spans 25 evals and 7 of 7 behavior components. Its strongest relative result is Enkrypt AI Safety Leaderboard (harmful_attack_non_success_rate, #7 of 260); its weakest is MORU (score, #12 of 13).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗claude-opus-4-7
- OpenRouter ↗anthropic/claude-opus-4.7
- System card ↗Exact model document · Anthropic · first party
- Release source ↗direct research preview identity date
- www.anthropic.com ↗
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 | #27 / 311 | 0.3618 | ↓ lower | Source ↗official | |
| AgentAbstainabstain | #1 / 17 | 79 | ↑ higher | Source ↗official | |
| AgentAbstaincar | #1 / 17 | 77.6 | ↑ higher | Source ↗official | |
| AgentAbstainpaired | #2 / 17 | 59.4 | ↑ higher | Source ↗official | |
| ANIMAscore | #4 / 19 | 0.735 | ↑ higher | Source ↗official | |
| BioSecBench-Refusalbalanced_refusal_score | #3 / 10 | 0.4259 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #6 / 105 | 0.785 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #5 / 43 | 44.5 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #3 / 48 | 19.2 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #4 / 49 | 28.1 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #19 / 45 | 85.7 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #23 / 51 | 17.1 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #16 / 32 | 48.2 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #39 / 48 | 21.8 | ↓ lower | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #7 / 105 | 2.642 | ↓ lower | Source ↗official | |
| DystopiaBenchbasaglia_score | #4 / 50 | 26.03 | ↓ lower | Source ↗official | |
| DystopiaBenchbaudrillard_score | #5 / 50 | 27.5 | ↓ lower | Source ↗official | |
| DystopiaBenchhuxley_score | #4 / 50 | 18.33 | ↓ lower | Source ↗official | |
| DystopiaBenchlaguardia_score | #2 / 50 | 30.07 | ↓ lower | Source ↗official | |
| DystopiaBenchorwell_score | #3 / 50 | 22.73 | ↓ lower | Source ↗official | |
| DystopiaBenchpetrov_score | #4 / 50 | 26.77 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #68 / 260 | 23.51 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #12 / 260 | 96.33 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #7 / 260 | 99.44 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #56 / 258 | 98.55 | ↑ higher | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #24 / 54 | 27.19 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #19 / 50 | 85.7 | ↓ lower | Source ↗official | |
| MANTAAWMS | #1 / 7 | 0.579 | ↑ higher | Source ↗official | |
| MANTAAWVS | #1 / 7 | 0.76 | ↑ higher | Source ↗official | |
| MORUscore | #12 / 13 | 67.2 | ↑ higher | Source ↗official | |
| ODCV-Benchaverage_severity | #1 / 12 | 0.0125 | ↓ lower | Source ↗official | |
| ODCV-Benchmisalignment_rate | #1 / 12 | 0 | ↓ lower | Source ↗official | |
| RefusalBenchyouden_j | #12 / 19 | 0.234 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #37 / 73 | 81.46 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #15 / 73 | 89.29 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #17 / 73 | 97.38 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #52 / 73 | 57.87 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #8 / 73 | 91.8 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #33 / 102 | 6.88 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #53 / 102 | 4.57 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #29 / 102 | 4.65 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #49 / 102 | 1.98 | ↓ lower | Source ↗official | |
| ToolPrivacyBenchprivate_mt_poi | #2 / 9 | 20.31 | ↓ lower | Source ↗official | |
| ToolPrivacyBenchpublic_mt_poi | #4 / 9 | 17.83 | ↓ lower | Source ↗official | |
| UK AISI active safety-research compromise continuationactive_compromise_continuation_rate_pct | #2 / 5 | 0.8 | ↓ lower | Source ↗official | |
| VETO Misfired Alignmentmisfired_alignment_rate_pct | #20 / 25 | 10.7 | ↓ 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.