AI Safety Index
62.86 / 100
14th of 17 published models
16 of 16 benchmarks · 100.0% of configured weight
Benchmark profile
Out of 100; higher is safer.
| Benchmark | Score visualization | Score out of 100 |
|---|---|---|
| SIM-VAILMental & Emotional Safety | 97.03 / 100 | |
| Spiral-BenchMental & Emotional Safety | 59.64 / 100 | |
| KORAYouth Safety | 50.33 / 100 | |
| PatientSafetyBenchMedical Advice Safety | 93.43 / 100 | |
| HealthBench-HardMedical Advice Safety | 26.73 / 100 | |
| ELEPHANTManipulation | 53.45 / 100 | |
| DarkBenchManipulation | 42.72 / 100 | |
| HumanAgencyBench (Autonomy)Manipulation | 41.63 / 100 | |
| ASK — AI Scam KnowledgeSecurity | 67.61 / 100 | |
| FairMT-BenchBias & Fairness | 55.25 / 100 | |
| ConfAIdePrivacy / Confidentiality | 98.00 / 100 | |
| SimpleQA VerifiedMisinformation | 73.96 / 100 | |
| SYCON-BenchMisinformation | 98.40 / 100 | |
| HumanAgencyBench (Correct Misinformation)Misinformation | 69.40 / 100 | |
| SystemCheck / RealGuardrailsRule Following | 90.79 / 100 | |
| AgentIFRule Following | 62.55 / 100 |
These are evaluations of API model configurations. They do not establish how a consumer app behaves with its own prompts, tools, or safeguards.
Exact configuration: model gemini-3.8-flash · configuration ID google.gemini-3.8-flash.medium-27b3898c4d83
