Model profile
Gemma 4 31B It
Evidence summary
Gemma 4 31B It has an estimated overall rank of #55; its 90% source-sensitivity interval is #35–#150. Its behavior-only rank is #65; company governance moves the combined estimate to #55. Published evidence spans 14 evals and 7 of 7 behavior components. Its strongest relative result is SM-Bench (adversarial, #2 of 78); its weakest is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #234 of 241).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gemma-4-31b
- Hugging Face ↗google/gemma-4-31B-it
- NVIDIA ↗qc69jvmznzxy/gemma-4-31b-it
- OpenRouter ↗google/gemma-4-31b-it
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
- huggingface.co ↗
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 | #167 / 327 | ↓0.8195 | Source ↗official | |
| Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating | #44 / 112 | ↑1442.0 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #60 / 106 | ↑0.225 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #218 / 241 | ↑8.53 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #234 / 241 | ↑66.67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #35 / 241 | ↑93.89 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #135 / 239 | ↑95.64 | Source ↗official | |
| JuICE Cultural-Error Span Detectionf1 | #5 / 10 | ↑0.445 | Source ↗official | |
| KIDBench Implicit Child Cueimplicit_child_cue_total_mean | #4 / 13 | ↑4.14 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #54 / 66 | ↑0.3522 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #30 / 70 | ↑0.7662 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #17 / 70 | ↑0.9651 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #24 / 67 | ↑0.6122 | Source ↗official | |
| RealityTest — Text AI-Identity Disclosuredisclosure_probability | #8 / 17 | ↑0.313 | Source ↗official | |
| SM-Benchadversarial | #2 / 78 | ↑91.71 | Source ↗official | |
| SM-Benchambiguous_interpretation | #11 / 78 | ↑90.77 | Source ↗official | |
| SM-Benchanti_hallucination | #22 / 78 | ↑96.86 | Source ↗official | |
| SM-Bencheq_boundaries | #29 / 78 | ↑67.7 | Source ↗official | |
| SM-Benchoverfit | #8 / 78 | ↑91.8 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #18 / 105 | ↑6.98 | Source ↗official | |
| SpeciEvalland_animal_4ns | #43 / 105 | ↓4.47 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #65 / 105 | ↓4.85 | Source ↗official | |
| SpeciEvalspeciesism | #97 / 105 | ↓2.88 | Source ↗official | |
| StereoTales Harmful Associationsbenign_significant_association_score | #5 / 23 | ↑88.1 | Source ↗official | |
| TACbase_welfare_rate | #50 / 74 | ↑23.72 | Source ↗self run | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #31 / 94 | ↑92.6 | Source ↗official | |
| Vigil Mental Health Safetyoverall_score | #12 / 23 | ↑46 | 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.