Model profile
Gemini 3.1 Flash Lite
Evidence summary
Gemini 3.1 Flash Lite has an estimated overall rank of #125; its 90% source-sensitivity interval is #55–#186. Its behavior-only rank is #140; company governance moves the combined estimate to #125. 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 MANTA (AWVS, #7 of 7).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gemini-3-1-flash-lite-preview
- OpenRouter ↗google/gemini-3.1-flash-lite
- 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 | #168 / 311 | 0.8163 | ↓ lower | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #84 / 105 | 0.11 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #7 / 43 | 45.4 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #39 / 48 | 89.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #39 / 49 | 72.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #40 / 45 | 95.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #45 / 51 | 55.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #23 / 32 | 51.9 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #14 / 48 | 15.6 | ↓ lower | Source ↗official | |
| DystopiaBenchbasaglia_score | #40 / 50 | 69.4 | ↓ lower | Source ↗official | |
| DystopiaBenchbaudrillard_score | #35 / 50 | 65.2 | ↓ lower | Source ↗official | |
| DystopiaBenchhuxley_score | #44 / 50 | 80.73 | ↓ lower | Source ↗official | |
| DystopiaBenchlaguardia_score | #41 / 50 | 69.77 | ↓ lower | Source ↗official | |
| DystopiaBenchorwell_score | #39 / 50 | 73.97 | ↓ lower | Source ↗official | |
| DystopiaBenchpetrov_score | #40 / 50 | 78.43 | ↓ lower | Source ↗official | |
| FORTRESSaverage_risk_score | #36 / 49 | 51.05 | ↓ lower | Source ↗official | |
| FORTRESSover_refusal_score | #11 / 46 | 2.15 | ↓ lower | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #42 / 50 | 95.8 | ↓ lower | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #10 / 31 | 7.633 | ↓ lower | Source ↗self run | |
| MANTAAWMS | #5 / 7 | 0.401 | ↑ higher | Source ↗official | |
| MANTAAWVS | #7 / 7 | 0.309 | ↑ higher | Source ↗official | |
| MASKlying_probability_pct | #45 / 53 | 51.6 | ↓ lower | Source ↗official | |
| PHAREbias_resistance_diagnostic | #37 / 66 | 0.4457 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #19 / 70 | 0.8057 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #13 / 70 | 0.9694 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #37 / 67 | 0.4772 | ↑ higher | Source ↗official | |
| RefusalBenchyouden_j | #10 / 19 | 0.4533 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #2 / 73 | 91.71 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #45 / 73 | 82.44 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #27 / 73 | 94.76 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #30 / 73 | 66.85 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #2 / 73 | 97.81 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #1 / 102 | 7 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #41 / 102 | 4.46 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #49 / 102 | 4.75 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #99 / 102 | 3.225 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #26 / 68 | 30.13 | ↑ higher | Source ↗self run | |
| VETO Misfired Alignmentmisfired_alignment_rate_pct | #13 / 25 | 8.3 | ↓ lower | Source ↗official | |
| Vigil Mental Health Safetyoverall_score | #13 / 23 | 38 | ↑ 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.