Mistral Medium 3.5 · No reasoning

Data published

Mistral Medium 3.5 · No reasoning

Reasoning
No reasoning
Made by
Mistral AI

No reasoning setting was sent; Mistral Medium 3.5 takes only none or high and has no medium. Mistral documents that a reply with reasoning carries a thinking chunk and a reply without it is a plain string; all 1,000 SimpleQA Verified replies (run 2026-09-21T224535.202567Z-275ae17e) are plain strings, so this configuration ran without reasoning.

AI Safety Index

37.25 / 100

17th of 17 published models

10 of 10 benchmarks · 100.0% of configured weight

Benchmark profile

Out of 100; higher is safer.

Benchmark scores for Mistral Medium 3.5 · No reasoning
BenchmarkScore visualizationScore out of 100
SIM-VAILMental & Emotional Safety66.79 / 100
Spiral-BenchMental & Emotional Safety39.26 / 100
KORAYouth Safety37.65 / 100
HealthBench-HardMedical Advice Safety15.13 / 100
ELEPHANTManipulation17.98 / 100
DarkBenchManipulation36.51 / 100
HumanAgencyBench (Autonomy)Manipulation29.61 / 100
ASK — AI Scam KnowledgeSecurity29.28 / 100
FairMT-BenchBias & Fairness56.56 / 100
SimpleQA VerifiedMisinformation21.95 / 100
HumanAgencyBench (Correct Misinformation)Misinformation15.16 / 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-medium-3-5 · configuration ID mistral.mistral-medium-3-5-c0d66b584183

Mistral Medium 3.5 — No reasoning Safety Benchmarks — Prosaic Intelligence