Model profile
Minimax M2.5
MiniMaxdeveloper
2026-02-12release date
#144 / 267overall rank
9eval lineages
Evidence summary
Minimax M2.5 has an estimated overall rank of #144; its 90% source-sensitivity interval is #69–#218. Published evidence spans 9 evals and 7 of 7 behavior components. Its strongest relative result is LiveSecBench (factuality, #5 of 43); its weakest is SM-Bench (adversarial, #67 of 73).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗minimax-m2-5
- Hugging Face ↗MiniMaxAI/MiniMax-M2.5
- OpenRouter ↗minimax/minimax-m2.5
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
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-eval | Rank | Value | Distribution | Better | Source |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #247 / 311 | 0.893 | ↓ lower | Source ↗official | |
| AgentAbstainabstain | #14 / 17 | 50.1 | ↑ higher | Source ↗official | |
| AgentAbstaincar | #14 / 17 | 49.6 | ↑ higher | Source ↗official | |
| AgentAbstainpaired | #11 / 17 | 41.9 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #89 / 105 | 0.085 | ↑ higher | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #24 / 105 | 9.521 | ↓ lower | Source ↗official | |
| DystopiaBenchbasaglia_score | #11 / 50 | 44.73 | ↓ lower | Source ↗official | |
| DystopiaBenchbaudrillard_score | #8 / 50 | 31.17 | ↓ lower | Source ↗official | |
| DystopiaBenchhuxley_score | #18 / 50 | 54.33 | ↓ lower | Source ↗official | |
| DystopiaBenchlaguardia_score | #22 / 50 | 63.47 | ↓ lower | Source ↗official | |
| DystopiaBenchorwell_score | #14 / 50 | 45.1 | ↓ lower | Source ↗official | |
| DystopiaBenchpetrov_score | #19 / 50 | 70.3 | ↓ lower | Source ↗official | |
| LiveSecBenchethics | #10 / 43 | 72.04 | ↑ higher | Source ↗official | |
| LiveSecBenchfactuality | #5 / 43 | 76.82 | ↑ higher | Source ↗official | |
| LiveSecBenchlegality | #21 / 43 | 48.6 | ↑ higher | Source ↗official | |
| LiveSecBenchprivacy | #19 / 43 | 59.81 | ↑ higher | Source ↗official | |
| LiveSecBenchpsychological_health | #22 / 43 | 50.99 | ↑ higher | Source ↗official | |
| SABERoverall_safety_rate | #6 / 13 | 26.34 | ↑ higher | Source ↗official | |
| SABERscenario_a_safety_rate | #4 / 13 | 32.83 | ↑ higher | Source ↗official | |
| SABERscenario_b_safety_rate | #5 / 13 | 34.83 | ↑ higher | Source ↗official | |
| SABERscenario_c_safety_rate | #11 / 13 | 12.23 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #67 / 73 | 73.17 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #48 / 73 | 82.14 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #66 / 73 | 77.49 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #44 / 73 | 61.24 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #56 / 73 | 50.82 | ↑ higher | Source ↗official | |
| TACbase_welfare_rate | #55 / 68 | 20.51 | ↑ higher | Source ↗self run |
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.