Model profile
Ministral 3 3B
Evidence summary
Safety. Ministral 3 3B has an estimated Safety rank of #119; its 90% source-sensitivity interval is #39–#322. Its behavior-only rank is #98; company governance moves the combined estimate to #119. Published Safety evidence spans 3 eval lineages and 3 of 7 components. Its strongest relative result is TAC (base_welfare_rate, #20 of 87); its weakest is Cisco AI Defense Rolling Single-Turn Leaderboard (single_turn_attack_success_rate, #82 of 104).
Freedom. Ministral 3 3B has an estimated Freedom rank of #76; its 90% source-sensitivity interval is #44–#479. Published Freedom evidence spans 2 eval lineages and 1 of 1 components. Its strongest relative result is SpeechMap model completion (complete_pct, #23 of 181); its weakest is Cisco AI Defense Rolling Single-Turn Leaderboard (single_turn_attack_success_rate, #23 of 104).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗ministral-3-3b
- Hugging Face ↗mistralai/Ministral-3-3B-Instruct-2512
- OpenRouter ↗mistralai/ministral-3b-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.
- jwest33/ministral-3-3b-instruct-2512-null-space-abliterated ↗Direct non-quantized finetune documented by Hugging Face metadata
- MuXodious/Ministral-3-3B-Instruct-2512-ARA-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 | #165 / 345 | ↓0.8019 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #82 / 104 | ↓64.21 | Source ↗official | |
| TACbase_welfare_rate | #20 / 87 | ↑35.26 | Source ↗self run |
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 | #23 / 104 | ↑64.21 | Source ↗official | |
| SpeechMap model completioncomplete_pct | #23 / 181 | ↑85.1 | Source ↗official |