Model profile
Llama 4 Maverick
Metadeveloper
2025-04-05release date
#177 / 312overall rank
31eval lineages
Evidence summary
Llama 4 Maverick has an estimated overall rank of #177; its 90% source-sensitivity interval is #116–#233. Its behavior-only rank is #172; company governance moves the combined estimate to #177. Published evidence spans 31 evals and 7 of 7 behavior components. Its strongest relative result is Anthropic Agentic Misalignment — blackmail (misaligned_action_rate_pct, #1 of 16); its weakest is SpeciesismBench (explicit_speciesism_scale, #7 of 7).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗llama-4-maverick
- Hugging Face ↗meta-llama/Llama-4-Maverick-17B-128E-Instruct
- OpenRouter ↗meta-llama/llama-4-maverick
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- 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 | #241 / 330 | ↓0.8888 | Source ↗official | |
| AgentDrive Safety Compliancescr | #30 / 48 | ↑75 | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #37 / 80 | ↑0.686 | Source ↗official | |
| Alignment Leaderboardcorrigibility | #15 / 24 | ↑4.175 | Source ↗official | |
| Alignment Leaderboardhonesty | #14 / 24 | ↑3.581 | Source ↗official | |
| Alignment Leaderboardnon_manipulation | #17 / 24 | ↑3.084 | Source ↗official | |
| Alignment Leaderboardrobustness | #9 / 24 | ↑3.867 | Source ↗official | |
| Alignment Leaderboardsafety | #18 / 24 | ↑3.567 | Source ↗official | |
| Alignment Leaderboardscheming | #19 / 24 | ↑3.422 | Source ↗official | |
| Anthropic Agentic Misalignment — blackmailmisaligned_action_rate_pct | #1 / 16 | ↓0 | Source ↗official | |
| Anthropic Agentic Misalignment — corporate espionagemisaligned_action_rate_pct | #7 / 16 | ↓44 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #54 / 106 | ↑0.28 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #72 / 104 | ↓56.23 | Source ↗official | |
| Confabulationsconfabulation_rate | #34 / 52 | ↓28.22 | Source ↗official | |
| DystopiaBenchbasaglia_score | #27 / 50 | ↓64.93 | Source ↗official | |
| DystopiaBenchbaudrillard_score | #36 / 50 | ↓65.83 | Source ↗official | |
| DystopiaBenchhuxley_score | #27 / 50 | ↓72.8 | Source ↗official | |
| DystopiaBenchlaguardia_score | #25 / 50 | ↓66.73 | Source ↗official | |
| DystopiaBenchorwell_score | #23 / 50 | ↓68.6 | Source ↗official | |
| DystopiaBenchpetrov_score | #18 / 50 | ↓68.63 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #204 / 241 | ↑10.34 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #62 / 241 | ↑91.17 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #77 / 241 | ↑82.22 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #82 / 239 | ↑97.82 | Source ↗official | |
| FORTRESSaverage_risk_score | #28 / 49 | ↓40.09 | Source ↗official | |
| FORTRESSover_refusal_score | #26 / 48 | ↓4.95 | Source ↗official | |
| HELM Safetyanthropic_red_team | #54 / 80 | ↑0.982 | Source ↗official | |
| HELM Safetybbq | #43 / 80 | ↑0.93 | Source ↗official | |
| HELM Safetyharmbench | #52 / 80 | ↑0.661 | Source ↗official | |
| HELM Safetysimple_safety_tests | #29 / 80 | ↑0.993 | Source ↗official | |
| HELM Safetyxstest | #32 / 80 | ↑0.965 | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #50 / 54 | ↑23.02 | Source ↗official | |
| Humanity's Last Exam RMS calibration error (Scale Labs)calibrationError | #32 / 39 | ↓83 | Source ↗official | |
| LiveSecBenchethics | #37 / 43 | ↑22.55 | Source ↗official | |
| LiveSecBenchfactuality | #35 / 43 | ↑29.74 | Source ↗official | |
| LiveSecBenchlegality | #30 / 43 | ↑27.05 | Source ↗official | |
| LiveSecBenchprivacy | #35 / 43 | ↑26.17 | Source ↗official | |
| LiveSecBenchpsychological_health | #29 / 43 | ↑35.37 | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #15 / 33 | ↓8.333 | Source ↗self run | |
| MASKlying_probability_pct | #41 / 53 | ↓50.27 | Source ↗official | |
| MT-JailBench CrescendoXsafety_score | #10 / 21 | ↑22.64 | Source ↗official | |
| ODCV-Benchaverage_severity | #10 / 12 | ↓1.682 | Source ↗official | |
| ODCV-Benchmisalignment_rate | #2 / 12 | ↓16.22 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #5 / 66 | ↑0.7365 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #43 / 70 | ↑0.7147 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #53 / 70 | ↑0.8925 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #34 / 67 | ↑0.4899 | Source ↗official | |
| Shelleducation_jsr | #10 / 14 | ↓0.696 | Source ↗official | |
| Shellfinance_jsr | #11 / 14 | ↓0.716 | Source ↗official | |
| Shellmanagement_jsr | #11 / 14 | ↓0.844 | Source ↗official | |
| SOSBenchbiology_pvr | #7 / 23 | ↓0.288 | Source ↗official | |
| SOSBenchchemistry_pvr | #7 / 23 | ↓0.238 | Source ↗official | |
| SOSBenchmedicine_pvr | #6 / 23 | ↓0.426 | Source ↗official | |
| SOSBenchpharmacology_pvr | #8 / 23 | ↓0.652 | Source ↗official | |
| SOSBenchphysics_pvr | #8 / 23 | ↓0.24 | Source ↗official | |
| SOSBenchpsychology_pvr | #6 / 23 | ↓0.242 | Source ↗official | |
| SpeciesismBenchexplicit_speciesism_scale | #7 / 7 | ↓3.3 | Source ↗official | |
| SpeciesismBenchmorally_wrong_rate | #2 / 8 | ↑42.67 | Source ↗official | |
| SpeciesismBenchspeciesism_recognition_rate | #4 / 8 | ↑88.97 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #22 / 113 | ↑6.97 | Source ↗official | |
| SpeciEvalland_animal_4ns | #77 / 113 | ↓4.72 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #79 / 113 | ↓4.9 | Source ↗official | |
| SpeciEvalspeciesism | #97 / 113 | ↓2.65 | Source ↗official | |
| TrustLLM contemporary collapsed applicationtrustllm | #7 / 8 | ↑0.6 | Source ↗official | |
| TukaBenchafri_jbb_cultural_asr | #4 / 6 | ↓22.4 | Source ↗official | |
| TukaBenchafri_jbb_harm_asr | #4 / 6 | ↓16.9 | Source ↗official | |
| TukaBenchafrijail_mono_asr | #4 / 6 | ↓22 | Source ↗official | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #36 / 94 | ↑91.8 | Source ↗official | |
| Vigil Mental Health Safetyoverall_score | #19 / 23 | ↑28 | 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.