Model profile
Gemini 2.5 Flash Lite
Googledeveloper
2025-06-17release date
#194 / 267overall rank
18eval lineages
Evidence summary
Gemini 2.5 Flash Lite has an estimated overall rank of #194; its 90% source-sensitivity interval is #48–#233. Its behavior-only rank is #203; company governance moves the combined estimate to #194. Published evidence spans 18 evals and 7 of 7 behavior components. Its strongest relative result is HELM Safety (xstest, #13 of 80); its weakest is CAIS Risk Index (agent_red_teaming, #43 of 43).
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
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 | #118 / 311 | 0.7542 | ↓ lower | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #43 / 80 | 0.658 | ↑ higher | Source ↗official | |
| ANIMAscore | #15 / 19 | 0.5838 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #43 / 43 | 96.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #47 / 48 | 96.3 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #48 / 49 | 83.4 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #45 / 45 | 109.4 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #44 / 51 | 54 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #42 / 48 | 22.8 | ↓ lower | Source ↗official | |
| Google Gemini 2.5 Flash-Lite Model Cardtext_safety_delta | #2 / 2 | 5.7 | ↓ lower | Source ↗official | |
| HELM Safetyanthropic_red_team | #44 / 80 | 0.987 | ↑ higher | Source ↗official | |
| HELM Safetybbq | #28 / 80 | 0.949 | ↑ higher | Source ↗official | |
| HELM Safetyharmbench | #49 / 80 | 0.67 | ↑ higher | Source ↗official | |
| HELM Safetysimple_safety_tests | #64 / 80 | 0.965 | ↑ higher | Source ↗official | |
| HELM Safetyxstest | #13 / 80 | 0.978 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #50 / 50 | 109.4 | ↓ lower | Source ↗official | |
| MORUscore | #8 / 13 | 71.34 | ↑ higher | Source ↗official | |
| PHAREbias_resistance_diagnostic | #34 / 66 | 0.4553 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #50 / 70 | 0.6809 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #66 / 70 | 0.7915 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #45 / 67 | 0.4284 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #43 / 102 | 6.83 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #81 / 102 | 4.81 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #19 / 102 | 4.5 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #29 / 102 | 1.74 | ↓ lower | Source ↗official |