Model profile
Gemini 2.5 Flash Lite
Evidence summary
Safety. Gemini 2.5 Flash Lite has an estimated Safety rank of #228; its 90% source-sensitivity interval is #94–#272. Its behavior-only rank is #239; company governance moves the combined estimate to #228. Published Safety evidence spans 23 eval lineages and 7 of 7 components. Its strongest relative result is Vectara HHEM Factual Consistency (factual_consistency_rate, #3 of 94); its weakest is CAIS Risk Index (agent_red_teaming, #49 of 49).
Freedom. Gemini 2.5 Flash Lite has an estimated Freedom rank of #48; its 90% source-sensitivity interval is #47–#224. Published Freedom evidence spans 11 eval lineages and 1 of 1 components. Its strongest relative result is CAIS Risk Index (bioweapons_assistance, #2 of 54); its weakest is Google Gemini 2.5 Flash-Lite Model Card (text_safety_delta, #1 of 2).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gemini-2-5-flash-lite
- OpenRouter ↗google/gemini-2.5-flash-lite
- Official model documentation ↗Family-level model document · google · first party
- Release source ↗direct research preview identity date
Safety evals
Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #132 / 345 | ↓0.7171 | Source ↗official | |
| Adversarial Poetry — AILuminate Baseline and Poetry ASRbaseline_asr | #15 / 24 | ↓8.67 | Source ↗official | |
| Adversarial Poetry — AILuminate Baseline and Poetry ASRpoetry_asr | #15 / 24 | ↓64.77 | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #43 / 80 | ↑0.658 | Source ↗official | |
| ANIMAscore | #18 / 22 | ↑0.5838 | Source ↗official | |
| Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating | #16 / 111 | ↑1461.0 | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #49 / 49 | ↓96.8 | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #53 / 54 | ↓96.3 | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #54 / 55 | ↓83.4 | Source ↗official | |
| CAIS Risk Indexmachiavelli | #51 / 51 | ↓109.4 | Source ↗official | |
| CAIS Risk Indexmask | #50 / 57 | ↓54 | Source ↗official | |
| CAIS Risk Indextextquests_harm | #44 / 54 | ↓22.8 | Source ↗official | |
| DelusionEvaldelusional_prevalence_pct | #13 / 16 | ↓56 | Source ↗official | |
| DelusionEvaldiscourages_harm_prevalence_pct | #12 / 16 | ↑16.5 | Source ↗official | |
| DelusionEvalfacilitates_harm_prevalence_pct | #12 / 16 | ↓5.6 | Source ↗official | |
| DelusionEvalrelationship_prevalence_pct | #10 / 16 | ↓33.6 | Source ↗official | |
| DelusionEvalsycophancy_prevalence_pct | #13 / 16 | ↓35 | Source ↗official | |
| Google Gemini 2.5 Flash-Lite Model Cardtext_safety_delta | #2 / 2 | ↓5.7 | Source ↗official | |
| HELM Safetyanthropic_red_team | #44 / 80 | ↑0.987 | Source ↗official | |
| HELM Safetybbq | #28 / 80 | ↑0.949 | Source ↗official | |
| HELM Safetyharmbench | #49 / 80 | ↑0.67 | Source ↗official | |
| HELM Safetysimple_safety_tests | #64 / 80 | ↑0.965 | Source ↗official | |
| HELM Safetyxstest | #13 / 80 | ↑0.978 | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #50 / 50 | ↓109.4 | Source ↗official | |
| MORUscore | #8 / 13 | ↑71.34 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #34 / 66 | ↑0.4553 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #50 / 70 | ↑0.6809 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #66 / 70 | ↑0.7915 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #45 / 67 | ↑0.4284 | Source ↗official | |
| SimpleQA Verifiedf1_score | #13 / 13 | ↑11.1 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #53 / 123 | ↑6.83 | Source ↗official | |
| SpeciEvalland_animal_4ns | #100 / 123 | ↓4.81 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #27 / 123 | ↓4.5 | Source ↗official | |
| SpeciEvalspeciesism | #38 / 123 | ↓1.74 | Source ↗official | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #3 / 94 | ↑96.7 | Source ↗official |
Freedom evals
Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| Adversarial Poetry — AILuminate Baseline and Poetry ASRbaseline_asr | #10 / 24 | ↑8.67 | Source ↗official | |
| Adversarial Poetry — AILuminate Baseline and Poetry ASRpoetry_asr | #10 / 24 | ↑64.77 | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #38 / 80 | ↓0.658 | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #2 / 54 | ↑96.3 | Source ↗official | |
| DelusionEvaldiscourages_harm_prevalence_pct | #5 / 16 | ↓16.5 | Source ↗official | |
| DelusionEvalfacilitates_harm_prevalence_pct | #5 / 16 | ↑5.6 | Source ↗official | |
| Google Gemini 2.5 Flash-Lite Model Cardtext_safety_delta | #1 / 2 | ↑5.7 | Source ↗official | |
| HELM Safetyanthropic_red_team | #35 / 80 | ↓0.987 | Source ↗official | |
| HELM Safetyharmbench | #32 / 80 | ↓0.67 | Source ↗official | |
| HELM Safetysimple_safety_tests | #17 / 80 | ↓0.965 | Source ↗official | |
| HELM Safetyxstest | #13 / 80 | ↑0.978 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #5 / 70 | ↓0.7915 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #23 / 67 | ↓0.4284 | Source ↗official | |
| SpeechMap model completioncomplete_pct | #26 / 181 | ↑84.6 | Source ↗official |