The LLM benchmark aggregator.

Kimi K2.6

Moonshot · 1T · open weights

Strongest in Computer use (#9 of 68, on 1 of its 2 boards), weakest in Harm refusal (#242 of 292). Above par in 13 of 26 scopes. Among the models it meets almost everywhere, it finishes behind Claude Fable 5 and Gemini 3.7 Flash and ahead of MiniMax M3 and Claude 3.7 Sonnet.

§ 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 preference62.5−6.3#51/3541/1

Knowledge53.3−9.2#55/1971/3

Academic knowledge57.9−6#25/1231/1

Agents54.1−12.3#69/2983/6

Computer use62.3−5.8#9/681/2

Knowledge work50.8−19.9#92/2082/2

Professional51.5−10.9#79/1941/2

Finance55−7.4#49/1621/1

Legal49.6−16.3#93/1731/1

Medical48.4−18.1#99/1491/1

Reasoning50−15.4#101/1952/5

Science59.1−4.4#31/1221/1

Mathematics43.5−20.3#115/1521/2

Coding47.6−18.6#103/1792/6

Agentic coding47.5−19.4#99/1732/5

Core abilities46.3−19#132/1991/4

Language46.1−20.5#42/621/1

Instruction following43.6−30.2#44/621/1

Safety48.1−13.3#231/3292/3

Secure code60.2−6#64/2701/1

Factual grounding47.8−22.9#68/1041/1

Fairness54.2−15.8#75/2921/2

Toxicity avoidance52.1−6.9#175/2681/1

Jailbreak resistance33−32#239/2681/1

Harm refusal44.8−16.7#242/2921/2

§ 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 Kimi K2.6, left for the other.

Mistral Medium 3.5won 16 · lost 0

Gemini 3.5 Flash-Litewon 17 · lost 1

Inkling Smallwon 14 · lost 2

GPT-5.4 miniwon 16 · lost 3

Magistral Small 1.2won 14 · lost 3

Command R+won 13 · lost 3

Gemini 1.5 Flashwon 13 · lost 3

GPT-4o miniwon 17 · lost 4

GPT-4.1 nanowon 17 · lost 4

Grok 4.1 Fastwon 17 · lost 4

Grok 4.3won 18 · lost 5

Ling 3.0 Flashwon 14 · lost 4

Mistral Large 3won 14 · lost 4

Command Awon 17 · lost 5

Gemini 1.5 Prowon 13 · lost 4

Llama 4 Scout Instructwon 16 · lost 5

Claude 3.5 Haikuwon 15 · lost 5

Gemini 2.0 Flashwon 15 · lost 5

DeepSeek V3won 17 · lost 6

Llama 3.3 70B Instructwon 14 · lost 5

Gemini 3.1 Flash-Litewon 13 · lost 5

Qwen3.6 Pluswon 13 · lost 5

NVIDIA Nemotron 3 Ultra 550B A55Bwon 17 · lost 7

GPT-5.4 nanowon 14 · lost 6

Llama 4 Maverick Instructwon 16 · lost 7

GPT OSS 120Bwon 16 · lost 7

GPT-4.1 miniwon 15 · lost 7

Claude Haiku 4.5won 15 · lost 7

DeepSeek R1won 15 · lost 7

Gemma 4 31Bwon 13 · lost 7

Gemini 2.5 Flash-Litewon 11 · lost 6

MiniMax M2.7won 14 · lost 8

Qwen3.8 27Bwon 12 · lost 7

MiniMax M2.5won 10 · lost 6

Grok 4 Fastwon 10 · lost 6

Kimi K2.5won 15 · lost 9

O3 Miniwon 13 · lost 8

Grok 3won 13 · lost 8

MiMo V2.5 Prowon 13 · lost 8

DeepSeek V3.2won 14 · lost 9

GLM-4.6won 14 · lost 9

GPT OSS 20Bwon 12 · lost 8

GPT-4owon 13 · lost 9

GLM-4.7won 10 · lost 7

O4 Miniwon 12 · lost 9

Grok 4won 12 · lost 9

GPT-6 Lunawon 13 · lost 10

GLM-5won 9 · lost 7

Inklingwon 14 · lost 11

Claude 3.5 Sonnetwon 10 · lost 8

GLM-4.5won 10 · lost 8

Claude Sonnet 4won 13 · lost 11

Claude 3.7 Sonnetwon 11 · lost 10

MiniMax M3won 13 · lost 12

GPT-5 nanowon 10 · lost 10

GPT-5.1won 11 · lost 11

Claude Sonnet 4.5won 12 · lost 12

GPT-5 miniwon 11 · lost 12

Claude Opus 4won 10 · lost 11

Claude Haiku 5.5won 8 · lost 9

DeepSeek V4 Flashwon 11 · lost 13

Kimi K2won 10 · lost 12

Gemini 2.5 Prowon 10 · lost 13

Gemini 2.5 Flashwon 9 · lost 12

GLM-5.1won 9 · lost 12

O1won 8 · lost 11

GPT-4.1won 9 · lost 13

Mistral Large 4won 9 · lost 13

O3won 9 · lost 13

Kimi K2 Thinkingwon 7 · lost 11

GLM-5.2won 9 · lost 15

Claude Opus 4.5won 7 · lost 12

Qwen3 235B A22B Instructwon 8 · lost 14

GLM-5.3 Flashwon 9 · lost 16

Grok 3 Miniwon 6 · lost 11

GPT-5won 8 · lost 15

GPT-5.6 Lunawon 8 · lost 16

Claude Sonnet 4.6won 8 · lost 16

GPT-6 Solwon 7 · lost 16

Claude Sonnet 5won 6 · lost 16

MiMo V2.6 Flashwon 5 · lost 14

Gemini 3.1 Prowon 5 · lost 15

Qwen3.7 Maxwon 6 · lost 18

Gemini 3.6 Flashwon 6 · lost 18

Claude Opus 4.6won 5 · lost 16

Grok 4.7won 5 · lost 17

GPT-5.2won 5 · lost 18

Gemini 3.5 Flashwon 5 · lost 19

MiMo V2.6 Prowon 4 · lost 16

DeepSeek V4 Prowon 5 · lost 20

GLM-5.3won 3 · lost 15

GPT-5.6 Terrawon 4 · lost 21

GPT-5.4won 4 · lost 21

Qwen3.8 Maxwon 4 · lost 21

GPT-6 Astrawon 3 · lost 16

Gemini 3 Prowon 3 · lost 16

Gemini 4 Argonwon 2 · lost 14

Grok 4.5won 3 · lost 21

Grok 4.6won 3 · lost 22

GPT-5.6 Solwon 3 · lost 22

Gemini 3.8 Flashwon 3 · lost 22

Claude Opus 4.7won 3 · lost 22

GPT-6.1 Solwon 2 · lost 15

Muse Spark 1.1won 2 · lost 23

Claude Sonnet 5.5won 1 · lost 16

DeepSeek V4.1 Flashwon 1 · lost 18

Kimi K3won 1 · lost 18

Claude Fable 5.1won 1 · lost 18

Claude Opus 4.8won 1 · lost 23

Claude Opus 5.5won 1 · lost 23

GPT-5.5won 1 · lost 24

Claude Opus 5won 1 · lost 24

Muse Spark 1.2won 0 · lost 17

Gemini 3.7 Flashwon 0 · lost 18

Claude Fable 5won 0 · lost 19

§ 3 · Sources

Where the numbers come from

8 publications, 33 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-10

LMArena Text 1461

GDPval-AA ↗1 measureread 2026-11-10

GDPval-AA 1047

AA-Briefcase ↗1 measureread 2026-11-10

AA-Briefcase 820

LiveBench 1.0 ↗8 measuresread 2026-11-10

LiveBench 1.0 70.5LiveBench 1.0 · Reasoning 79.4LiveBench 1.0 · Coding 78.6LiveBench 1.0 · Agentic Coding 46.9LiveBench 1.0 · Mathematics 84.3LiveBench 1.0 · Data Analysis 65.1LiveBench 1.0 · Language 75.1LiveBench 1.0 · Instruction Following 64.4

OSWorld ↗1 measureread 2026-11-10

OSWorld 73.1%

Vals.ai ↗14 measuresread 2026-11-10

Vals · Legal Research Bench 15.87%Vals · CaseLaw 61.2%Vals · LegalBench 84.74%Vals · Finance Agent 44.9%Vals · CorpFin 66.74%Vals · TaxEval 74.65%Vals · MortgageTax 65.82%Vals · MedCode 40.14%Vals · MedScribe 78.15%Vals · SWE-bench Verified 76.2% (Mini-SWE-agent)Vals · Vibe Code Bench 37.89% (OpenHands)Vals · Code Migration 27.77%Vals · GPQA Diamond 89.14%Vals · MMLU Pro 87.57%

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

Enkrypt · Jailbreak risk 27.1%Enkrypt · Harmful content risk 9.4%Enkrypt · CBRN risk 48.5%Enkrypt · Toxicity risk 4.7%Enkrypt · Bias risk 71.3%Enkrypt · Insecure code risk 12.4%

Vectara Hallucination Leaderboard ↗1 measureread 2026-11-10

Vectara · Factual consistency 89.2%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=kimi-k2-6)](https://publicai.io/model-index/m/kimi-k2-6)