AI Safety Evidence

← All models

Developer

Mistral AI

40 indexed models; 21 currently meet the evidence threshold for the overall ranking. Together they have results from 68 evaluations.

Official Mistral AI model information ↗

Company governance evidence

Mistral AI is represented at -1.20 SD relative to the matched Future of Life Institute edition. Provenance: Direct. This company-level evidence contributes 10% of overall model rank.

Models by Mistral AI

ModelEvalsComponentsRankRelease date
Mistral Large66/7932024-02-26
Ministral 3 3B33/71002025-10-31
Mistral Medium45/71792023-12-11
Mistral Small 3.152/72142025-03-11
Mistral Medium 3.575/72312026-03-31
Mistral Small 476/72362026-01-23
Magistral Medium86/72442025-06-10
Mistral Medium 3.197/72752025-08-12
Mistral Small 396/72792025-01-30
Mistral 7B Instruct247/72812023-09-27
Mistral Small 3.245/72822025-06-19
Mistral Nemo55/72912024-07-18
Mixtral 8x22B Instruct95/73042024-04-17
Mistral Medium 365/73062025-05-07
Mistral 7B44/73082023-09-27
Mistral Large 3166/73212025-12-01
Mixtral 8x7B Instruct136/73272023-12-10
Mistral Large 2145/73322024-07-24
Magistral Small55/73422025-06-04
Ministral 3 14B32/73452025-10-31
Ministral 3 8B42/73462025-10-31
Devstral11/7—2025-05-21
Devstral 222/7—2025-12-09
Devstral Small11/7—2025-05-12
Devstral Small 211/7—2025-12-09
Ministral 3B11/7—2024-10-16
Ministral 8B24/7—2024-10-16
Ministral 8B Moderated14/7—2024-10-16
Mistral Large Latest13/7——
Mistral Large Moderated14/7—2024-02-26
Mistral Nemo Minitron 8B Instruct13/7—2024-10-02
Mistral Small12/7—2024-09-17
Mistral Small 211/7——
Mistral Small Latest13/7——
Mixtral 8x7B Base11/7—2023-12-01
Open Codestral Mamba13/7—2024-07-16
Pixtral 12B11/7—2024-09-11
Pixtral Large 250211/7—2025-04-08
Voxtral Mini 3B 250711/7—2025-07-01
Voxtral Small 24B 250722/7—2025-07-01
Evaluations covering Mistral AI models (68)

AA-Omniscience · AbstentionBench · Adversarial Humanities Benchmark (AHB) — Table 5 · Adversarial Poetry — AILuminate Baseline and Poetry ASR · Adversarial Robustness · AgentDrive Safety Compliance · AgentHarm · AILuminate General Purpose AI Chat · AIRBench 2024 Safety Scenarios · Alignment Leaderboard · AnimalHarmBench · Arena Factuality — Text Arena (factuality-only weighting) · AuAu Authoritarian Response Audit · BioTIER · BlueBench AttaQ-100 · BullshitBench v2 · CAIS Risk Index · CASE-Bench · ChineseSafe · Cisco AI Defense Rolling Single-Turn Leaderboard · COMPL-AI AI-Identity Disclosure · COMPL-AI LLM RuLES Multi-Turn Rule Following · COMPL-AI TensorTrust Goal-Hijacking Resistance · Concordia AI Risk Monitor · Confabulations · Contextual MoralChoice · DSPSafeBench · DystopiaBench · Emergent Collusion · Enkrypt AI Safety Leaderboard · FlagEval Safety and Values · FORTRESS · HalluVerse-M3 Hallucination Recognition · HarmBench · HELM Safety · HUMAINE Trust, Ethics and Safety · Human Pathogen Capabilities Test (HPCT) — overall refusal · Humanity's Last Exam RMS calibration error (Scale Labs) · JailBench · Large-scale Moral Machine experiment on LLMs · LiveSecBench · MANTA · MASK · MedSafetyBench · Microsoft Phi Safety Panels · MT-JailBench CrescendoX · NESSiE Necessary Safety Benchmark · Opposite-Narrator Sycophancy · OR-Bench · PacifAIst · PHARE · Qwen2 Safety Panel · RealityTest — Text AI-Identity Disclosure · RefusalBench · S-Eval · SafeDialBench · SALAD-Bench · Shell · SM-Bench · SORRY-Bench · SpeciEval · StereoTales Harmful Associations · SuperCLUE Safety · TAC · UAVBench safety-critical decision recognition · Vectara HHEM Factual Consistency · VETO Misfired Alignment · XSTest