The LLM benchmark aggregator.

Llama 4 Maverick Instruct

Meta · 402B · open weights

Strongest in Safe-prompt compliance (#33 of 82), weakest in Agents (#220 of 268). Above par in 8 of 27 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 4.8 and Grok 4.7 and ahead of Mistral Large and Gemini 2.0 Flash.

§ 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.

Core abilities51.1−16.3#80/2041/3

General intelligence51.6−18.5#82/2041/3

Knowledge48.1−15.4#97/1381/2

Academic knowledge47.7−16.2#89/1231/1

Safety52.4−9.1#121/3373/3

Safe-prompt compliance55.3−5.1#33/821/1

Factual grounding54.5−16.1#40/1011/1

Jailbreak resistance59.8−5.1#50/2721/1

Toxicity avoidance55.8−3.2#89/2721/1

Harm refusal53.2−8.4#103/3002/2

Fairness47.6−22.4#149/3002/2

Secure code41.8−24.7#207/2741/1

Coding43.8−25.3#125/1651/5

Agentic coding42.5−26.2#125/1571/4

Reasoning40.3−26.6#152/1783/4

Science45.4−18.1#90/1221/1

Mathematics39.1−26.6#115/1401/2

Reasoning38.6−30.5#119/1402/3

Professional41.3−20.3#153/1681/1

Legal49.7−17.4#80/1511/1

Finance44.7−18.2#115/1511/1

Medical31.4−35.4#138/1401/1

Human preference50−17.5#192/3421/1

Agents41.9−26.2#220/2682/5

Tool use49.6−24.4#36/811/1

Knowledge work32.8−40.8#169/1781/2

§ 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 Llama 4 Maverick Instruct, left for the other.

Gemini 1.5 Flashwon 15 · lost 2

Llama 4 Scout Instructwon 21 · lost 4

Command R Pluswon 15 · lost 3

Mistral Smallwon 16 · lost 5

GPT-4o Miniwon 18 · lost 6

GPT-4.1 Nanowon 19 · lost 7

Mistral Medium 3.5won 12 · lost 5

Command Awon 16 · lost 8

Gemini 1.5 Prowon 12 · lost 7

GPT OSS 20Bwon 13 · lost 9

Gemini 2.0 Flashwon 13 · lost 10

Mistral Largewon 10 · lost 9

Claude 3.5 Haikuwon 10 · lost 10

GPT OSS 120Bwon 12 · lost 14

Mistral Large 3won 9 · lost 11

Grok 4.3won 10 · lost 14

GLM-4.6won 10 · lost 15

Ling 3.0 Flashwon 7 · lost 11

GPT-4owon 8 · lost 14

Gemma 4 31Bwon 8 · lost 14

MiniMax M3won 9 · lost 16

Kimi K2.5won 9 · lost 16

O3 Miniwon 8 · lost 15

Grok 4.1 Fastwon 8 · lost 15

DeepSeek V3won 9 · lost 17

Gemini 3.5 Flash Litewon 6 · lost 12

Claude Haiku 4.5won 9 · lost 18

Kimi K2.6won 8 · lost 17

Kimi K2won 8 · lost 18

Gemini 2.5 Flash Litewon 6 · lost 14

GPT-5.4 Nanowon 6 · lost 14

DeepSeek V4 Flashwon 7 · lost 17

Gemini 3.6 Flashwon 7 · lost 17

DeepSeek R1won 7 · lost 18

GLM-5.2won 7 · lost 18

GPT-4.1 Miniwon 7 · lost 19

DeepSeek V3.2won 7 · lost 19

GLM-4.5won 5 · lost 14

Gemini 3.1 Flash Litewon 5 · lost 14

GPT-5.4 Miniwon 5 · lost 14

Mimo V2.6 Prowon 5 · lost 14

Mimo V2.5 Prowon 6 · lost 17

Claude 3.5 Sonnetwon 5 · lost 15

GPT-5 Nanowon 6 · lost 18

GLM-5.3 Flashwon 6 · lost 18

Qwen3.7 Maxwon 6 · lost 18

DeepSeek V4 Prowon 6 · lost 19

Grok 3won 5 · lost 16

GPT-6 Lunawon 5 · lost 16

GPT-4.1won 6 · lost 20

Grok 4 Fastwon 4 · lost 14

Qwen3.6 Pluswon 4 · lost 14

Gemini 3.5 Flashwon 5 · lost 19

Kimi K2 Thinkingwon 4 · lost 16

GPT-6 Solwon 4 · lost 18

GLM-5.1won 4 · lost 18

Claude Sonnet 4.6won 4 · lost 19

GLM-5won 3 · lost 15

Gemini 3.8 Flashwon 4 · lost 20

Grok 4won 4 · lost 21

Inklingwon 4 · lost 21

GPT-5.6 Lunawon 4 · lost 21

Gemini 2.5 Flashwon 4 · lost 22

Gemini 2.5 Prowon 4 · lost 22

GPT-5 Miniwon 4 · lost 23

MiniMax M2.7won 3 · lost 18

Gemini 3 Prowon 3 · lost 20

Grok 4.6won 3 · lost 21

GPT-5.1won 3 · lost 21

Grok 4.5won 3 · lost 21

Qwen3.8 Maxwon 3 · lost 21

Muse Spark 1.1won 3 · lost 21

Qwen3 235B A22B Instructwon 3 · lost 22

GPT-5.4won 3 · lost 22

GPT-5.5won 3 · lost 22

GPT-5.6 Solwon 3 · lost 22

GPT-6 Astrawon 2 · lost 15

O4 Miniwon 3 · lost 23

Claude Opus 4.5won 2 · lost 18

Gemini 3.1 Prowon 2 · lost 18

Claude Opus 4.6won 2 · lost 19

Claude 3.7 Sonnetwon 2 · lost 20

Claude Opus 5won 2 · lost 22

GPT-5.2won 2 · lost 23

GPT-5.6 Terrawon 2 · lost 23

GPT-5won 2 · lost 24

Claude Sonnet 4.5won 2 · lost 25

O1won 1 · lost 20

Claude Opus 5.5won 1 · lost 20

Claude Opus 4won 1 · lost 23

O3won 1 · lost 24

Claude Sonnet 5won 1 · lost 24

Claude Opus 4.7won 1 · lost 24

Claude Sonnet 4won 1 · lost 25

DeepSeek V4.1 Flashwon 0 · lost 17

Muse Spark 1.2won 0 · lost 17

GLM-5.3won 0 · lost 18

Gemini 3.7 Flashwon 0 · lost 18

Kimi K3won 0 · lost 18

Claude Fable 5.1won 0 · lost 18

Claude Fable 5won 0 · lost 18

Qwen3.8 27Bwon 0 · lost 19

Grok 4.7won 0 · lost 21

Claude Opus 4.8won 0 · lost 24

§ 3 · Sources

Where the numbers come from

11 publications, 29 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-09-26

LMArena Text 1327

GDPval-AA ↗1 measureread 2026-09-26

GDPval-AA -291

ARC-AGI-2 ↗1 measureread 2026-09-26

ARC-AGI-2 0%

Aider polyglot ↗1 measureread 2026-09-26

Aider polyglot 15.6%

BFCL ↗1 measureread 2026-09-26

BFCL v4 37.29%

Kagi LLM Benchmark ↗1 measureread 2026-09-26

Kagi LLM Benchmark 55.9%

SimpleBench ↗1 measureread 2026-09-26

SimpleBench 27.7%

Vals.ai ↗10 measuresread 2026-09-26

Vals · LegalBench 77.81%Vals · CorpFin 49.73%Vals · TaxEval 66.56%Vals · MortgageTax 58.51%Vals · MedQA 43.3%Vals · MedCode 36.51%Vals · MedScribe 54.22%Vals · GPQA Diamond 69.44%Vals · MMLU Pro 79.42%Vals · AIME 25.21%

HELM Safety ↗5 measuresread 2026-09-26

HELM Safety · HarmBench 66.1%HELM Safety · SimpleSafetyTests 99.3%HELM Safety · Anthropic Red Team 98.2%HELM Safety · BBQ 93%HELM Safety · XSTest 96.5%

Enkrypt AI Safety Leaderboard ↗6 measuresread 2026-09-26

Enkrypt · Jailbreak risk 4.5%Enkrypt · Harmful content risk 17.8%Enkrypt · CBRN risk 8.8%Enkrypt · Toxicity risk 2.2%Enkrypt · Bias risk 89.7%Enkrypt · Insecure code risk 50.2%

Vectara Hallucination Leaderboard ↗1 measureread 2026-09-26

Vectara · Factual consistency 91.8%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=llama-4-maverick-instruct)](https://publicai.io/model-index/m/llama-4-maverick-instruct)