Model profile
Claude 3.7 Sonnet
Anthropicdeveloper
2025-02-24release date
#26 / 312overall rank
27eval lineages
Evidence summary
Claude 3.7 Sonnet has an estimated overall rank of #26; its 90% source-sensitivity interval is #12–#98. Its behavior-only rank is #30; company governance moves the combined estimate to #26. Published evidence spans 27 evals and 7 of 7 behavior components. Its strongest relative result is HELM Safety (simple_safety_tests, #1 of 80); its weakest is Anthropic Claude 4 System Card (benign_request_refusal, #3 of 3).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗claude-3-7-sonnet
- OpenRouter ↗anthropic/claude-3.7-sonnet
- 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 | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #42 / 330 | ↓0.3997 | Source ↗official | |
| AgentDojotargeted_attack_success_rate | #3 / 15 | ↓0.0731 | Source ↗official | |
| AgentDojoutility_under_attack | #1 / 15 | ↑0.7727 | Source ↗official | |
| AgentDrive Safety Compliancescr | #32 / 48 | ↑68.75 | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #19 / 80 | ↑0.818 | Source ↗official | |
| Anthropic Agentic Misalignment — blackmailmisaligned_action_rate_pct | #6 / 16 | ↓65 | Source ↗official | |
| Anthropic Agentic Misalignment — corporate espionagemisaligned_action_rate_pct | #1 / 16 | ↓4 | Source ↗official | |
| Anthropic Agentic Misalignment — lethal actionmisaligned_action_rate_pct | #1 / 10 | ↓0 | Source ↗official | |
| Anthropic Claude 4 System Cardagentic_coding_safety | #3 / 3 | ↑0.85 | Source ↗official | |
| Anthropic Claude 4 System Cardbenign_request_refusal | #3 / 3 | ↓0.0045 | Source ↗official | |
| Anthropic Claude 4 System Cardharmful_request_safety | #2 / 3 | ↑0.9896 | Source ↗official | |
| Anthropic Claude 4 System Cardstrongreject_jailbreak_success | #3 / 3 | ↓0.0809 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #27 / 106 | ↑0.46 | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #43 / 51 | ↓75 | Source ↗official | |
| CAIS Risk Indexmask | #23 / 53 | ↓16.9 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #27 / 104 | ↓11.27 | Source ↗official | |
| Confabulationsconfabulation_rate | #17 / 52 | ↓16.58 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #117 / 241 | ↑16.02 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #17 / 241 | ↑95.17 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #17 / 241 | ↑98.33 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #31 / 239 | ↑99.32 | Source ↗official | |
| FORTRESSaverage_risk_score | #27 / 49 | ↓38.01 | Source ↗official | |
| FORTRESSover_refusal_score | #24 / 48 | ↓4.45 | Source ↗official | |
| HELM Safetyanthropic_red_team | #10 / 80 | ↑0.997 | Source ↗official | |
| HELM Safetybbq | #47 / 80 | ↑0.921 | Source ↗official | |
| HELM Safetyharmbench | #30 / 80 | ↑0.843 | Source ↗official | |
| HELM Safetysimple_safety_tests | #1 / 80 | ↑1 | Source ↗official | |
| HELM Safetyxstest | #33 / 80 | ↑0.964 | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #49 / 54 | ↑23.36 | Source ↗official | |
| Humanity's Last Exam RMS calibration error (Scale Labs)calibrationError | #26 / 39 | ↓80 | Source ↗official | |
| LLM Ethics Benchmarkscore | #1 / 5 | ↑90.9 | Source ↗official | |
| MASKlying_probability_pct | #18 / 53 | ↓24.07 | Source ↗official | |
| OpenAgentSafetyllm_judge_safety_vulnerable | #2 / 7 | ↓51.2 | Source ↗official | |
| OpenAgentSafetyrule_based_safety_vulnerable | #5 / 7 | ↓32.85 | Source ↗official | |
| OpenAgentSafetysuccessful_completion | #3 / 7 | ↑33.88 | Source ↗official | |
| PandaBench JBB direct-request panelsafety_rate | #1 / 46 | ↑1 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #56 / 66 | ↑0.3377 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #9 / 70 | ↑0.8517 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #21 / 70 | ↑0.9552 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #22 / 67 | ↑0.6345 | Source ↗official | |
| Reward Hacking Benchmarkintegrity_score | #7 / 13 | ↑96.1 | Source ↗official | |
| SOSBenchbiology_pvr | #5 / 23 | ↓0.229 | Source ↗official | |
| SOSBenchchemistry_pvr | #5 / 23 | ↓0.208 | Source ↗official | |
| SOSBenchmedicine_pvr | #4 / 23 | ↓0.35 | Source ↗official | |
| SOSBenchpharmacology_pvr | #6 / 23 | ↓0.579 | Source ↗official | |
| SOSBenchphysics_pvr | #5 / 23 | ↓0.171 | Source ↗official | |
| SOSBenchpsychology_pvr | #4 / 23 | ↓0.168 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #85 / 113 | ↑6.53 | Source ↗official | |
| SpeciEvalland_animal_4ns | #32 / 113 | ↓4.35 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #17 / 113 | ↓4.47 | Source ↗official | |
| SpeciEvalspeciesism | #74 / 113 | ↓2.19 | Source ↗official | |
| SYCON Benchfalse_presupposition_tof | #3 / 11 | ↑2.92 | Source ↗official | |
| SYCON Benchunethical_queries_tof | #1 / 11 | ↑2.73 | Source ↗official |