Model profile
MiniMax M2
Evidence summary
Safety. MiniMax M2 has an estimated Safety rank of #131; its 90% source-sensitivity interval is #62–#267. Its behavior-only rank is #134; company governance moves the combined estimate to #131. Published Safety evidence spans 7 eval lineages and 6 of 7 components. Its strongest relative result is Concordia — Shutdown-Resistance (safety_score, #1 of 53); its weakest is SpeciEval (sea_animal_4ns, #115 of 123).
Freedom. MiniMax M2 has an estimated Freedom rank of #499; its 90% source-sensitivity interval is #366–#545. Published Freedom evidence spans 5 eval lineages and 1 of 1 components. Its strongest relative result is Concordia — FRT-SciKnowEval-BiologicalHarmfulQA (safety_score, #11 of 45); its weakest is Concordia — Fortress-Privacy/Scams (safety_score, #48 of 54).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗minimax-m2
- Hugging Face ↗MiniMaxAI/MiniMax-M2
- OpenRouter ↗minimax/minimax-m2
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
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 | #281 / 345 | ↓0.909 | Source ↗official | |
| Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating | #95 / 111 | ↑1405.0 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #32 / 104 | ↓18.05 | Source ↗official | |
| Concordia — Agentic-Misalignmentsafety_score | #23 / 54 | ↑82.5 | Source ↗official | |
| Concordia — AirBench-Deceptionsafety_score | #19 / 63 | ↑90.74 | Source ↗official | |
| Concordia — AirBench-Manipulationsafety_score | #12 / 56 | ↑98 | Source ↗official | |
| Concordia — AirBench-SecurityRiskssafety_score | #18 / 63 | ↑98.2 | Source ↗official | |
| Concordia — APEsafety_score | #13 / 55 | ↑58.36 | Source ↗official | |
| Concordia — CyberSecEval2-PromptInjectionsafety_score | #51 / 63 | ↑73.31 | Source ↗official | |
| Concordia — DarkBenchsafety_score | #10 / 55 | ↑66.98 | Source ↗official | |
| Concordia — Fortress-Biologicalsafety_score | #12 / 54 | ↑81.11 | Source ↗official | |
| Concordia — Fortress-Chemicalsafety_score | #14 / 54 | ↑79.07 | Source ↗official | |
| Concordia — Fortress-Privacy/Scamssafety_score | #7 / 54 | ↑79.81 | Source ↗official | |
| Concordia — FRT-AirBench-Manipulationsafety_score | #11 / 45 | ↑50 | Source ↗official | |
| Concordia — FRT-AirBench-SecurityRiskssafety_score | #14 / 45 | ↑38.67 | Source ↗official | |
| Concordia — FRT-SciKnowEval-BiologicalHarmfulQAsafety_score | #33 / 45 | ↑1.667 | Source ↗official | |
| Concordia — FRT-SOSBench-Chemsafety_score | #12 / 45 | ↑65 | Source ↗official | |
| Concordia — MASKsafety_score | #14 / 62 | ↑80.51 | Source ↗official | |
| Concordia — SciKnowEval-BiologicalHarmfulQAsafety_score | #32 / 63 | ↑82.49 | Source ↗official | |
| Concordia — SciKnowEval-ChemicalHarmfulQAsafety_score | #18 / 63 | ↑32.82 | Source ↗official | |
| Concordia — Shutdown-Resistancesafety_score | #1 / 53 | ↑100 | Source ↗official | |
| Concordia — SOSBench-Biosafety_score | #27 / 63 | ↑87.6 | Source ↗official | |
| Concordia — SOSBench-Chemsafety_score | #33 / 63 | ↑86.8 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #15 / 248 | ↑51.94 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #155 / 248 | ↑86 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #57 / 248 | ↑89.44 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #36 / 246 | ↑99.09 | Source ↗official | |
| LiveSecBenchethics | #12 / 43 | ↑69.21 | Source ↗official | |
| LiveSecBenchfactuality | #13 / 43 | ↑62.29 | Source ↗official | |
| LiveSecBenchlegality | #12 / 43 | ↑67.96 | Source ↗official | |
| LiveSecBenchprivacy | #14 / 43 | ↑64.5 | Source ↗official | |
| LiveSecBenchpsychological_health | #10 / 43 | ↑69.47 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #59 / 123 | ↑6.82 | Source ↗official | |
| SpeciEvalland_animal_4ns | #110 / 123 | ↓4.97 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #115 / 123 | ↓5.2 | Source ↗official | |
| SpeciEvalspeciesism | #44 / 123 | ↓1.81 | 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.
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.
UGI Political Values
Taiwan Sovereignty Benchmark Pro
| Dimension | Value | Distribution |
|---|---|---|
| Pro-Taiwan rubric compatibility | 55 | |
| Warning-phrase rate | 35 | |
| Soft-censorship rate | 0 | |
| API-error rate | 0 |