Model profile
Ministral 3 8B
Evidence summary
Safety. Ministral 3 8B has an estimated Safety rank of #333; its 90% source-sensitivity interval is #322–#333. Its behavior-only rank is #331; company governance moves the combined estimate to #333. Published Safety evidence spans 4 eval lineages and 2 of 7 components. Its strongest relative result is Cisco AI Defense Rolling Single-Turn Leaderboard (single_turn_attack_success_rate, #89 of 104); its weakest is MT-JailBench CrescendoX (safety_score, #21 of 21).
Freedom. Ministral 3 8B has an estimated Freedom rank of #11; its 90% source-sensitivity interval is #18–#127. Published Freedom evidence spans 3 eval lineages and 1 of 1 components. Its strongest relative result is MT-JailBench CrescendoX (safety_score, #1 of 21); its weakest is Cisco AI Defense Rolling Single-Turn Leaderboard (single_turn_attack_success_rate, #16 of 104).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗ministral-3-8b
- Hugging Face ↗mistralai/Ministral-3-8B-Instruct-2512
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
Finetunes
Direct finetunes linked to this canonical base model. Quantized and repackaged derivatives are excluded.
- MuXodious/Ministral-3-8B-Instruct-2512-PaperWitch-heresy ↗Direct non-quantized finetune documented by Hugging Face metadata
Safety evals
Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #320 / 345 | ↓0.9416 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #89 / 104 | ↓71.09 | Source ↗official | |
| MT-JailBench CrescendoXsafety_score | #21 / 21 | ↑0.63 | Source ↗official | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #91 / 94 | ↑78.3 | Source ↗official |
Freedom evals
Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #16 / 104 | ↑71.09 | Source ↗official | |
| MT-JailBench CrescendoXsafety_score | #1 / 21 | ↓0.63 | Source ↗official | |
| UGI Leaderboard — base-model willingnesswillingness_adherence_score | #24 / 156 | ↑6.333 | Source ↗official | |
| UGI Leaderboard — base-model willingnesswillingness_direct_score | #21 / 156 | ↑5 | Source ↗official |
Values evaluations
Descriptive values and political-framing results are separate from safety/ethics ranks. Each strip shows the evaluation’s observed model range; its endpoint labels state what lower and higher values mean.