‹ PublicAI Index
The LLM benchmark aggregator.
MiniMax M2.7
MiniMax · 229B · open weights
Strongest in Secure code (#6 of 270), weakest in Coding (#148 of 179). Above par in 11 of 22 scopes. Among the models it meets almost everywhere, it finishes behind Claude Fable 5 and Claude Fable 5.1 and ahead of Grok 4.1 Fast and DeepSeek R1.
§ 1 · Profile
What it is good at
Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.
Safety58.5−2.9#14/3292/3
Secure code65.8−0.4#6/2701/1
Fairness70leads#10/2921/2
Jailbreak resistance62.7−2.3#18/2681/1
Harm refusal56.3−5.2#43/2921/2
Toxicity avoidance56.4−2.6#59/2681/1
Factual grounding42.5−28.2#86/1041/1
Reasoning55.7−9.7#37/1951/5
Mathematics56.6−7.2#40/1521/2
Science57.3−6.2#42/1221/1
Human preference58.1−10.7#115/3541/1
Human preference58.1−10.7#115/3541/1
Knowledge49.6−12.9#118/1971/3
Academic knowledge49−14.9#83/1231/1
Agents49.5−16.9#135/2982/6
Knowledge work49.2−21.5#100/2082/2
Professional46.1−16.3#145/1941/2
Legal47.9−18#108/1731/1
Medical45.8−20.7#116/1491/1
Finance43.7−18.7#129/1621/1
Coding42.6−23.6#148/1791/6
Agentic coding42.2−24.7#143/1731/5
§ 2 · Head to head
What it beats, and what beats it
The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for MiniMax M2.7, left for the other.
§ 3 · Sources
Where the numbers come from
6 publications, 24 figures. Every one links to the page it was read from.
LMArena Text 1415
GDPval-AA 1016
AA-Briefcase 719
Vals · Legal Research Bench 10.58%Vals · CaseLaw 60.88%Vals · LegalBench 83.98%Vals · Finance Agent 27.89%Vals · CorpFin 61.19%Vals · TaxEval 66.56%Vals · MedCode 34.44%Vals · MedScribe 79.87%Vals · SWE-bench Verified 73.8% (Mini-SWE-agent)Vals · Vibe Code Bench 11.93% (OpenHands)Vals · Code Migration 8.69%Vals · GPQA Diamond 86.62%Vals · MMLU Pro 80.43%Vals · AIME 91.04%
Enkrypt · Jailbreak risk 2.2%Enkrypt · Harmful content risk 2.2%Enkrypt · CBRN risk 15%Enkrypt · Toxicity risk 1.8%Enkrypt · Bias risk 45.5%Enkrypt · Insecure code risk 0.9%
Vectara · Factual consistency 87.1%
Badge
[](https://publicai.io/model-index/m/minimax-m2-7)