Mistral Large 4 · No reasoning
- Reasoning
- No reasoning
- Made by
- Mistral AI
No reasoning setting was sent, and the model did not reason: Mistral marks a reasoning reply with a thinking chunk, and 38,493 of the 38,494 replies checked across 13 of the 14 benchmarks carry none (SIM-VAIL's transcripts were not checked). The exception is one ELEPHANT reply resubmitted on 2026-10-07 after a provider 504, after Mistral changed its default.
AI Safety Index
63.26 / 100
12th 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 | 96.04 / 100 | |
| Spiral-BenchMental & Emotional Safety | 69.17 / 100 | |
| KORAYouth Safety | 64.72 / 100 | |
| PatientSafetyBenchMedical Advice Safety | 97.47 / 100 | |
| HealthBench-HardMedical Advice Safety | 18.73 / 100 | |
| ELEPHANTManipulation | 32.15 / 100 | |
| DarkBenchManipulation | 81.66 / 100 | |
| HumanAgencyBench (Autonomy)Manipulation | 39.58 / 100 | |
| ASK — AI Scam KnowledgeSecurity | 55.95 / 100 | |
| FairMT-BenchBias & Fairness | 69.39 / 100 | |
| ConfAIdePrivacy / Confidentiality | 95.44 / 100 | |
| SimpleQA VerifiedMisinformation | 12.43 / 100 | |
| SYCON-BenchMisinformation | 78.20 / 100 | |
| HumanAgencyBench (Correct Misinformation)Misinformation | 41.86 / 100 | |
| SystemCheck / RealGuardrailsRule Following | 61.92 / 100 | |
| AgentIFRule Following | 59.20 / 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 mistral-large-4 · configuration ID mistral.mistral-large-4-8fb0931066d6
