← Models

Model profile

Magistral Small

Mistral AIdeveloper
2025-06-04release date
#252 / 267overall rank
3eval lineages

Evidence summary

Magistral Small has an estimated overall rank of #252; its 90% source-sensitivity interval is #133–#265. Its behavior-only rank is #245; company governance moves the combined estimate to #252. Published evidence spans 3 evals and 5 of 7 behavior components. Its strongest relative result is AA-Omniscience (hallucination_rate, #120 of 311); its weakest is PHARE (harm_resistance_diagnostic, #68 of 70).

Compare this model

Only models sharing at least one published sub-eval are listed.

Official and reference links

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-evalRankValueDistributionBetterSource
AA-Omnisciencehallucination_rate#120 / 3110.7548↓ lowerSource ↗official
Enkrypt AI Safety Leaderboardbias_attack_non_success_rate#141 / 26014.99↑ higherSource ↗official
Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate#231 / 26076.17↑ higherSource ↗official
Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate#241 / 26033.33↑ higherSource ↗official
Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate#204 / 25891.64↑ higherSource ↗official
PHAREbias_resistance_diagnostic#28 / 660.4802↑ higherSource ↗official
PHAREhallucination_resistance_diagnostic#64 / 700.6244↑ higherSource ↗official
PHAREharm_resistance_diagnostic#68 / 700.7623↑ higherSource ↗official
PHAREjailbreak_resistance_diagnostic#57 / 670.3709↑ higherSource ↗official