The LLM benchmark aggregator.

MiniMax M2.5

MiniMax · 229B · open weights

Strongest in Mathematics (#43 of 152, on 1 of its 2 boards), weakest in Safety (#159 of 329). Above par in 10 of 18 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 4.8 and Claude Fable 5 and ahead of Inkling and Gemini 3.1 Flash-Lite.

§ 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.

Reasoning51.6−13.8#87/1952/5

Mathematics56−7.8#43/1521/2

Science54.2−9.3#62/1221/1

Reasoning42.9−26.1#99/1471/3

Core abilities50.8−14.5#88/1991/4

General intelligence51.2−18.2#76/1781/3

Professional49.9−12.5#98/1941/2

Medical54.8−11.7#44/1491/1

Finance50−12.4#98/1621/1

Legal47−18.9#112/1731/1

Coding45.7−20.5#118/1791/6

Agentic coding45.5−21.4#114/1731/5

Knowledge49.4−13.1#120/1971/3

Academic knowledge48.6−15.3#84/1231/1

Human preference55.8−13#142/3541/1

Safety50.7−10.7#159/3291/3

Factual grounding52.1−18.6#45/1041/1

§ 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 M2.5, left for the other.

Mistral Small 3.1 24B Instructwon 11 · lost 0

Grok 2won 11 · lost 0

NVIDIA Nemotron 3.5 Lightning 30B A3Bwon 12 · lost 0

GPT OSS 20Bwon 14 · lost 0

GPT-4.1 nanowon 16 · lost 1

Command Awon 16 · lost 1

GPT-4o miniwon 15 · lost 1

Llama 4 Scout Instructwon 15 · lost 1

Gemini 1.5 Prowon 12 · lost 1

Command R+won 11 · lost 1

Gemini 1.5 Flashwon 10 · lost 1

Gemini 2.0 Flashwon 15 · lost 2

Mistral Medium 3.5won 13 · lost 2

Llama 3.3 70B Instructwon 11 · lost 2

Mistral Large 2.1won 10 · lost 2

Mistral Medium 3.1won 10 · lost 2

GPT OSS 120Bwon 15 · lost 3

GPT-4owon 13 · lost 3

GPT-5 nanowon 13 · lost 3

Claude Haiku 4.5won 13 · lost 3

Gemini 2.5 Flash-Litewon 12 · lost 3

Mistral Large 3won 14 · lost 4

Claude 3.5 Haikuwon 10 · lost 3

MiniMax M2.1won 12 · lost 4

O3 Miniwon 11 · lost 4

Llama 4 Maverick Instructwon 13 · lost 5

GPT-5.4 nanowon 13 · lost 5

DeepSeek V3won 12 · lost 6

Kimi K2won 12 · lost 6

GPT-4.1 miniwon 11 · lost 6

Claude 3.5 Sonnetwon 9 · lost 5

Gemma 4 31Bwon 9 · lost 5

DeepSeek V3.1won 7 · lost 4

NVIDIA Nemotron 3 Ultra 550B A55Bwon 10 · lost 7

GPT-5 miniwon 10 · lost 8

DeepSeek V3.2won 10 · lost 8

Ling 3.0 Flashwon 6 · lost 5

MiMo V2.5won 7 · lost 6

Grok 4.3won 8 · lost 7

DeepSeek R1won 9 · lost 8

GLM-4.6won 9 · lost 8

Gemini 3.1 Flash-Litewon 9 · lost 8

Inklingwon 9 · lost 8

Qwen3.6 Pluswon 7 · lost 7

Gemini 3.5 Flash-Litewon 8 · lost 8

GPT-5.4 miniwon 8 · lost 9

GPT-6 Lunawon 7 · lost 8

MiMo V2.6 Flashwon 6 · lost 7

Grok 4 Fastwon 8 · lost 10

Grok 3won 8 · lost 10

Qwen3 235B A22B Instructwon 8 · lost 10

Grok 4.1 Fastwon 7 · lost 9

Mistral Large 4won 6 · lost 8

MiMo V2.5 Prowon 6 · lost 8

Step 5won 5 · lost 7

Grok 3 Miniwon 7 · lost 10

MiniMax M3won 7 · lost 10

Claude 3.7 Sonnetwon 6 · lost 9

MiniMax M2.7won 6 · lost 9

Claude Sonnet 4won 7 · lost 11

GPT-4.1won 7 · lost 11

Qwen 3.5 Pluswon 5 · lost 8

O4 Miniwon 6 · lost 10

Inkling Smallwon 6 · lost 10

GLM-4.7won 6 · lost 10

Kimi K2.6won 6 · lost 10

Grok 4.20won 6 · lost 10

DeepSeek V4won 4 · lost 7

Claude Opus 4.1won 5 · lost 9

Kimi K2 Thinkingwon 5 · lost 10

Qwen3 Maxwon 5 · lost 10

Gemini 2.5 Flashwon 6 · lost 12

Kimi K2.5won 6 · lost 12

Qwen3.8 27Bwon 5 · lost 11

Qwen3.7 Maxwon 5 · lost 11

DeepSeek V4 Flashwon 5 · lost 12

GLM-5.3 Flashwon 5 · lost 12

Gemini 3.1 Prowon 5 · lost 12

GLM-4.5won 4 · lost 10

MiMo V2.6 Prowon 4 · lost 10

Claude Sonnet 4.5won 5 · lost 13

DeepSeek V4 Prowon 5 · lost 13

Claude Opus 4.5won 5 · lost 13

Qwen3.5 Flashwon 4 · lost 11

Gemini 3 Flashwon 4 · lost 12

GPT-5.6 Lunawon 4 · lost 13

Claude Sonnet 4.6won 4 · lost 13

GLM-5.2won 4 · lost 13

Claude Opus 4won 4 · lost 14

Gemini 2.5 Prowon 4 · lost 14

O1won 3 · lost 12

Grok 4won 3 · lost 14

O3won 3 · lost 14

Qwen3.8 Maxwon 3 · lost 14

Grok 4.7won 2 · lost 12

GLM-5.1won 2 · lost 13

Muse Spark 1.1won 2 · lost 13

Gemini 4 Argonwon 2 · lost 13

Claude Sonnet 5won 2 · lost 14

DeepSeek V4.1 Flashwon 2 · lost 14

Gemini 3.5 Flashwon 2 · lost 14

Gemini 3.6 Flashwon 2 · lost 14

Kimi K3won 2 · lost 14

Grok 4.6won 2 · lost 15

Grok 4.5won 2 · lost 15

GPT-5.6 Terrawon 2 · lost 15

Gemini 3.8 Flashwon 2 · lost 15

GLM-5won 2 · lost 16

GPT-5.6 Solwon 2 · lost 16

Claude Opus 4.6won 2 · lost 16

Muse Spark 1.3won 1 · lost 11

Claude Haiku 5.5won 1 · lost 13

Muse Spark 1.2won 1 · lost 13

GPT-6 Solwon 1 · lost 14

Gemini 3.7 Flashwon 1 · lost 14

GPT-5.1won 1 · lost 15

GLM-5.3won 1 · lost 15

Claude Opus 5.5won 1 · lost 15

GPT-5won 1 · lost 17

GPT-5.2won 1 · lost 17

GPT-5.4won 1 · lost 17

GPT-5.5won 1 · lost 17

Gemini 3 Prowon 1 · lost 17

Claude Opus 4.7won 1 · lost 17

Kimi K3.1 Prowon 0 · lost 11

Muse Wonderwon 0 · lost 12

Muse Sparkwon 0 · lost 13

Claude Sonnet 5.5won 0 · lost 14

GPT-6.1 Solwon 0 · lost 14

GPT-6 Astrawon 0 · lost 16

Claude Opus 5won 0 · lost 16

Claude Fable 5.1won 0 · lost 16

Claude Fable 5won 0 · lost 16

Claude Opus 4.8won 0 · lost 17

§ 3 · Sources

Where the numbers come from

5 publications, 14 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-10

LMArena Text 1391

ARC-AGI-2 ↗1 measureread 2026-11-10

ARC-AGI-2 4.9%

Kagi LLM Benchmark ↗1 measureread 2026-11-10

Kagi LLM Benchmark 55.2%

Vals.ai ↗10 measuresread 2026-11-10

Vals · CaseLaw 53.48%Vals · LegalBench 79.96%Vals · CorpFin 59.6%Vals · TaxEval 68.15%Vals · MedQA 92.53%Vals · SWE-bench Verified 74.2% (Mini-SWE-agent)Vals · Vibe Code Bench 14.85% (OpenHands)Vals · GPQA Diamond 82.07%Vals · MMLU Pro 80.09%Vals · AIME 88.75%

Vectara Hallucination Leaderboard ↗1 measureread 2026-11-10

Vectara · Factual consistency 90.9%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=minimax-m2-5)](https://publicai.io/model-index/m/minimax-m2-5)