Model profile
Gemini 2.5 Flash
Googledeveloper
2025-04-17release date
#121 / 267overall rank
31eval lineages
Evidence summary
Gemini 2.5 Flash has an estimated overall rank of #121; its 90% source-sensitivity interval is #44–#186. Its behavior-only rank is #131; company governance moves the combined estimate to #121. Published evidence spans 31 evals and 7 of 7 behavior components. Its strongest relative result is HELM Safety (xstest, #1 of 80); its weakest is Google Gemini 2.5 Flash Model Card (text_safety_delta, #2 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
- OpenRouter ↗google/gemini-2.5-flash
- 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 | #100 / 311 | 0.7066 | ↓ lower | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #36 / 80 | 0.687 | ↑ higher | Source ↗official | |
| Alignment Leaderboardcorrigibility | #13 / 24 | 4.204 | ↑ higher | Source ↗official | |
| Alignment Leaderboardhonesty | #11 / 24 | 3.613 | ↑ higher | Source ↗official | |
| Alignment Leaderboardnon_manipulation | #14 / 24 | 3.326 | ↑ higher | Source ↗official | |
| Alignment Leaderboardrobustness | #22 / 24 | 2.907 | ↑ higher | Source ↗official | |
| Alignment Leaderboardsafety | #6 / 24 | 4.038 | ↑ higher | Source ↗official | |
| Alignment Leaderboardscheming | #14 / 24 | 3.608 | ↑ higher | Source ↗official | |
| Anthropic Agentic Misalignment — blackmailmisaligned_action_rate_pct | #15 / 16 | 96 | ↓ lower | Source ↗official | |
| Anthropic Agentic Misalignment — corporate espionagemisaligned_action_rate_pct | #14 / 16 | 100 | ↓ lower | Source ↗official | |
| Anthropic Agentic Misalignment — lethal actionmisaligned_action_rate_pct | #6 / 10 | 83 | ↓ lower | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #67 / 105 | 0.19 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #39 / 43 | 91.9 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #46 / 48 | 95.9 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #44 / 49 | 80 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #26 / 45 | 89.4 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #40 / 51 | 50.9 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #7 / 48 | 11.7 | ↓ lower | Source ↗official | |
| Confabulationsconfabulation_rate | #6 / 52 | 4.455 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #65 / 260 | 24.84 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #191 / 260 | 84.17 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #180 / 260 | 54.63 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #206 / 258 | 91.48 | ↑ higher | Source ↗official | |
| Google Gemini 2.5 Flash Model Cardtext_safety_delta | #2 / 2 | 4.2 | ↓ lower | Source ↗official | |
| HELM Safetyanthropic_red_team | #41 / 80 | 0.988 | ↑ higher | Source ↗official | |
| HELM Safetybbq | #7 / 80 | 0.977 | ↑ higher | Source ↗official | |
| HELM Safetyharmbench | #60 / 80 | 0.626 | ↑ higher | Source ↗official | |
| HELM Safetysimple_safety_tests | #48 / 80 | 0.98 | ↑ higher | Source ↗official | |
| HELM Safetyxstest | #1 / 80 | 0.988 | ↑ higher | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #28 / 54 | 27.07 | ↑ higher | Source ↗official | |
| LiveSecBenchethics | #31 / 43 | 31.43 | ↑ higher | Source ↗official | |
| LiveSecBenchfactuality | #27 / 43 | 38.55 | ↑ higher | Source ↗official | |
| LiveSecBenchlegality | #17 / 43 | 56.07 | ↑ higher | Source ↗official | |
| LiveSecBenchprivacy | #29 / 43 | 31.6 | ↑ higher | Source ↗official | |
| LiveSecBenchpsychological_health | #19 / 43 | 54.27 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #26 / 50 | 89.4 | ↓ lower | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #20 / 31 | 8.833 | ↓ lower | Source ↗self run | |
| MASKlying_probability_pct | #43 / 53 | 50.87 | ↓ lower | Source ↗official | |
| PacifAIstp_score | #1 / 7 | 90.31 | ↑ higher | Source ↗official | |
| PHAREbias_resistance_diagnostic | #22 / 66 | 0.5174 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #40 / 70 | 0.7213 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #35 / 70 | 0.9366 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #58 / 67 | 0.3705 | ↑ higher | Source ↗official | |
| PropensityBenchscore | #11 / 14 | 68 | ↓ lower | Source ↗official | |
| Social Welfare Function Benchmarkfairness | #18 / 19 | 0.438 | ↑ higher | Source ↗official | |
| SOSBenchbiology_pvr | #9 / 23 | 0.336 | ↓ lower | Source ↗official | |
| SOSBenchchemistry_pvr | #10 / 23 | 0.338 | ↓ lower | Source ↗official | |
| SOSBenchmedicine_pvr | #7 / 23 | 0.462 | ↓ lower | Source ↗official | |
| SOSBenchpharmacology_pvr | #10 / 23 | 0.684 | ↓ lower | Source ↗official | |
| SOSBenchphysics_pvr | #10 / 23 | 0.424 | ↓ lower | Source ↗official | |
| SOSBenchpsychology_pvr | #9 / 23 | 0.326 | ↓ lower | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #66 / 102 | 6.67 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #96 / 102 | 5.05 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #53 / 102 | 4.78 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #75 / 102 | 2.39 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #8 / 68 | 39.74 | ↑ higher | Source ↗official | |
| UAVBench safety-critical decision recognitionethical_safety_critical_accuracy | #12 / 27 | 0.715 | ↑ higher | Source ↗official | |
| Vigil Mental Health Safetyoverall_score | #19 / 23 | 28 | ↑ higher | Source ↗official |