Model profile
Gemini 3.1 Pro Preview
Evidence summary
Gemini 3.1 Pro Preview has an estimated overall rank of #18; its 90% source-sensitivity interval is #6–#94. Its behavior-only rank is #19; company governance moves the combined estimate to #18. Published evidence spans 20 evals and 7 of 7 behavior components. Its strongest relative result is SpeciEval (belief_animal_sentience, #1 of 102); its weakest is MASK (lying_probability_pct, #52 of 53).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gemini-3-1-pro-preview
- OpenRouter ↗google/gemini-3.1-pro-preview
- Official model documentation ↗Family-level model document · google · 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 | Better | Source |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #50 / 311 | 0.4987 | ↓ lower | Source ↗official | |
| AgentAbstainabstain | #6 / 17 | 65.4 | ↑ higher | Source ↗official | |
| AgentAbstaincar | #3 / 17 | 65.7 | ↑ higher | Source ↗official | |
| AgentAbstainpaired | #1 / 17 | 59.5 | ↑ higher | Source ↗official | |
| BioSecBench-Refusalbalanced_refusal_score | #6 / 10 | 0.3931 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #43 / 105 | 0.34 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #16 / 43 | 68 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #28 / 48 | 73.5 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #19 / 49 | 50.3 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #25 / 45 | 89.2 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #43 / 51 | 53.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #11 / 32 | 43.3 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #28 / 48 | 18.8 | ↓ lower | Source ↗official | |
| FORTRESSaverage_risk_score | #24 / 49 | 29.76 | ↓ lower | Source ↗official | |
| FORTRESSover_refusal_score | #18 / 46 | 3.13 | ↓ lower | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #7 / 54 | 29.01 | ↑ higher | Source ↗official | |
| LiveSecBenchethics | #23 / 43 | 45.45 | ↑ higher | Source ↗official | |
| LiveSecBenchfactuality | #21 / 43 | 51.4 | ↑ higher | Source ↗official | |
| LiveSecBenchlegality | #19 / 43 | 50.36 | ↑ higher | Source ↗official | |
| LiveSecBenchprivacy | #12 / 43 | 67.16 | ↑ higher | Source ↗official | |
| LiveSecBenchpsychological_health | #6 / 43 | 76.44 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #25 / 50 | 89.2 | ↓ lower | Source ↗official | |
| MASKlying_probability_pct | #52 / 53 | 57.6 | ↓ lower | Source ↗official | |
| ODCV-Benchaverage_severity | #11 / 12 | 1.825 | ↓ lower | Source ↗official | |
| ODCV-Benchmisalignment_rate | #10 / 12 | 43.75 | ↓ lower | Source ↗official | |
| RefusalBenchyouden_j | #2 / 19 | 0.9217 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #4 / 73 | 90.73 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #35 / 73 | 85.71 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #7 / 73 | 98.95 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #24 / 73 | 68.26 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #17 / 73 | 84.15 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #1 / 102 | 7 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #27 / 102 | 4.35 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #41 / 102 | 4.7 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #97 / 102 | 3.03 | ↓ lower | Source ↗official | |
| VETO Misfired Alignmentmisfired_alignment_rate_pct | #2 / 25 | 4.9 | ↓ lower | Source ↗official | |
| Vigil Mental Health Safetyoverall_score | #10 / 23 | 48 | ↑ higher | 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.
UGI Political Values
Agent-ValueBench Moral Foundations (MFT08)
Agent-ValueBench HEXACO
| Dimension | Value | Distribution |
|---|---|---|
| Openness to experience | 4.1 | |
| Honesty-humility | 5.4 | |
| Extraversion | 5.9 | |
| Agreeableness | 5.7 | |
| Conscientiousness | 7.1 |