The LLM benchmark aggregator.

MiMo V2.6 Pro

Xiaomi · 1T · open weights

Strongest in Web research (#11 of 35, on 1 of its 2 boards), weakest in Fairness (#168 of 292). Above par in 22 of 28 scopes. Among the models it meets almost everywhere, it finishes behind Claude Fable 5.1 and Claude Opus 5 and ahead of Kimi K2 Thinking and MiniMax M2.7.

§ 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.

Human preference64.3−4.5#27/3541/1

Agents58−8.4#30/2983/6

Web research55−8.9#11/351/2

Long-horizon work49.5−24.5#15/311/1

Knowledge work65.5−5.2#16/2082/2

Computer use57−11.1#21/681/2

Professional54−8.4#42/1942/2

Cybersecurity57.1−6.8#14/471/1

Document design51.9−16.8#16/311/1

IT operations41.4−32.4#23/331/1

Legal56−9.9#33/1731/1

Medical56.1−10.4#33/1491/1

Finance56.2−6.2#35/1621/1

Coding53.7−12.5#51/1792/6

Agentic coding53.9−13#48/1732/5

Core abilities50.5−14.8#94/1992/4

Data analysis48.5−16.7#22/311/1

General intelligence51.6−17.8#73/1782/3

Reasoning47.8−17.6#115/1952/5

Mathematics53−10.8#71/1521/2

Reasoning42.1−26.9#109/1471/3

Safety52.1−9.3#128/3291/3

Jailbreak resistance57.9−7.1#68/2681/1

Toxicity avoidance55.2−3.8#95/2681/1

Secure code53.4−12.8#124/2701/1

Harm refusal51.4−10.1#134/2921/2

Fairness46−24#168/2921/2

§ 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 MiMo V2.6 Pro, left for the other.

Command Awon 19 · lost 0

Gemma 4 31Bwon 23 · lost 1

DeepSeek V3won 20 · lost 1

Step 5won 19 · lost 1

Qwen3.8 27Bwon 19 · lost 2

Llama 4 Scout Instructwon 17 · lost 2

Gemini 2.0 Flashwon 17 · lost 2

Grok 4.3won 17 · lost 2

GPT-4.1 nanowon 18 · lost 3

NVIDIA Nemotron 3 Ultra 550B A55Bwon 18 · lost 3

Kimi K2won 17 · lost 3

GLM-4.6won 17 · lost 3

Grok 3won 16 · lost 3

MiMo V2.5 Prowon 15 · lost 3

DeepSeek V4 Flashwon 18 · lost 4

GPT-4.1 miniwon 17 · lost 4

GLM-5.2won 17 · lost 4

Kimi K2.6won 16 · lost 4

GLM-5.3 Flashwon 22 · lost 6

Grok 4.1 Fastwon 13 · lost 4

DeepSeek V3.2won 16 · lost 5

Gemini 2.5 Flashwon 15 · lost 5

Qwen3.7 Maxwon 15 · lost 5

Mistral Large 4won 19 · lost 7

MiMo V2.6 Flashwon 16 · lost 6

GPT OSS 20Bwon 13 · lost 5

Llama 4 Maverick Instructwon 15 · lost 6

GPT OSS 120Bwon 15 · lost 6

DeepSeek R1won 15 · lost 6

GPT-6 Lunawon 20 · lost 8

Claude 3.5 Haikuwon 12 · lost 5

GPT-4o miniwon 14 · lost 6

GPT-4.1won 14 · lost 6

GPT-4owon 13 · lost 6

DeepSeek V4 Prowon 15 · lost 7

Grok 4.5won 16 · lost 8

Inklingwon 18 · lost 9

MiniMax M3won 15 · lost 8

Grok 4won 13 · lost 7

Qwen3 235B A22B Instructwon 13 · lost 7

Kimi K2.5won 14 · lost 8

O3 Miniwon 12 · lost 7

GLM-5.1won 12 · lost 7

Claude Sonnet 4.6won 12 · lost 7

Gemini 2.5 Prowon 13 · lost 8

Gemini 3.5 Flashwon 13 · lost 8

Gemini 3.6 Flashwon 13 · lost 8

Gemini 3.8 Flashwon 17 · lost 11

Claude 3.7 Sonnetwon 11 · lost 8

Claude Haiku 4.5won 11 · lost 8

GPT-5.6 Lunawon 13 · lost 10

MiniMax M2.7won 10 · lost 8

Kimi K2 Thinkingwon 9 · lost 8

GPT-5 nanowon 9 · lost 9

O4 Miniwon 10 · lost 10

Claude Sonnet 4won 11 · lost 11

Grok 4.6won 13 · lost 14

Qwen3.8 Maxwon 13 · lost 14

GPT-5.4won 10 · lost 11

Kimi K3won 10 · lost 11

Claude Opus 4won 9 · lost 10

GPT-5.1won 9 · lost 10

Gemini 3 Prowon 8 · lost 9

GPT-5.6 Terrawon 13 · lost 15

Claude Sonnet 4.5won 10 · lost 12

Claude Sonnet 5won 10 · lost 12

DeepSeek V4.1 Flashwon 10 · lost 13

GPT-5 miniwon 9 · lost 12

O3won 9 · lost 12

Claude Haiku 5.5won 9 · lost 13

GPT-6 Solwon 9 · lost 13

GPT-5won 8 · lost 13

Kimi K3.1 Prowon 7 · lost 12

GPT-6 Astrawon 7 · lost 16

GPT-5.2won 6 · lost 14

Muse Spark 1.1won 6 · lost 14

GLM-5.3won 5 · lost 12

Claude Opus 4.6won 5 · lost 12

Claude Sonnet 5.5won 5 · lost 17

Grok 4.7won 6 · lost 22

GPT-5.6 Solwon 5 · lost 19

Muse Spark 1.3won 4 · lost 17

GPT-5.5won 4 · lost 18

GPT-6.1 Solwon 4 · lost 18

Muse Wonderwon 3 · lost 17

Gemini 4 Argonwon 3 · lost 20

Claude Fable 5won 2 · lost 15

Claude Opus 5.5won 3 · lost 25

Claude Opus 4.8won 2 · lost 19

Claude Opus 4.7won 2 · lost 19

Claude Opus 5won 2 · lost 21

Claude Fable 5.1won 1 · lost 21

§ 3 · Sources

Where the numbers come from

7 publications, 28 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-10

LMArena Text 1480

Artificial Analysis ↗1 measureread 2026-11-10

Artificial Analysis Intelligence Index 30

GDPval-AA ↗1 measureread 2026-11-10

GDPval-AA 1692

AA-Briefcase ↗1 measureread 2026-11-10

AA-Briefcase 1512

LiveBench ↗8 measuresread 2026-11-10

LiveBench 66.1LiveBench · Agentic 36.4LiveBench · Reasoning 79.6LiveBench · Coding 64.5LiveBench · Browser Use 77.3LiveBench · Data Analysis 72.3LiveBench · Research 84.8LiveBench · Design 74.3

Vals.ai ↗10 measuresread 2026-11-10

Vals · Legal Research Bench 47.12%Vals · Harvey Legal Agent Benchmark 13.75%Vals · Finance Agent 57.34%Vals · MedCode 44.97%Vals · MedScribe 88.31%Vals · CyberBench 72.86%Vals · SRE Bench 3.05%Vals · Vibe Code Bench 85.22% (OpenHands)Vals · Code Migration 43.01%Vals · ProofBench 70%

Enkrypt AI Safety Leaderboard ↗6 measuresread 2026-11-10

Enkrypt · Jailbreak risk 6.2%Enkrypt · Harmful content risk 2.2%Enkrypt · CBRN risk 28.3%Enkrypt · Toxicity risk 2.6%Enkrypt · Bias risk 84%Enkrypt · Insecure code risk 26.2%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=mimo-v2-6-pro)](https://publicai.io/model-index/m/mimo-v2-6-pro)