llmboard.aiAI model intelligence
Home

Model Rankings

OverallOpen ModelsAgentCodingReasoningMathKnowledgeInstruction FollowingTextVision
Image GenerationImage Editing
Video GenerationImage to VideoVideo Editing
Text to SpeechSpeech to Text
Embeddings

Efficiency

Chat Token PricingImage PricingVideo PricingAudio Pricing
Chat Speed & LatencyProvider Reliability

Benchmarks

GPQAMMLU-ProAIME 2025SWE-Bench VerifiedMMLUHumanity's Last ExamLiveCodeBenchMATHHumanEvalMMMU-Pro
All Benchmarks

Tools

Model DirectoryCompare Models

Scoring & Data

Scoring & Data
393 models668 benchmarks

Leaderboard Center

Overall RankingCodingCore BenchmarksPrice & ValueRuntime Performance

Modalities

All ModelsImage GenerationImage EditingVideo GenerationImage-to-VideoVideo EditingText-to-SpeechSpeech-to-TextEmbeddings

Data & Methods

Scoring MethodAll BenchmarksReasoningMath

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai

math benchmark

MMMLU Leaderboard

Multilingual Massive Multitask Language Understanding dataset released by OpenAI, featuring professionally translated MMLU test questions across 14 languages including Arabic, Bengali, German, Spanish, French, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Swahili, Yoruba, and Chinese. Contains approximately 15,908 multiple-choice questions per language covering 57 subjects.

Updated Aug 17, 2026

Models49
Model coverage49
MetricScore
EvidenceB

On this page

  • Ranking
  • Highlights
  • Distribution
  • Top models
  • About
  • FAQ

MMMLU Ranking

Higher score ranks better on this benchmark.

30 of 49 rows
Columns

Show columns

Sort by
Rank
Model
Score
Percentile
Participants
Evidence
Evaluated
Rank01ModelANClaude Mythos PreviewAnthropicScore92.7%Percentile100.0%Participants49EvidenceCEvaluatedAug 17, 2026
Rank02ModelGOGemini 3.1 ProGoogleScore92.6%Percentile97.9%Participants49EvidenceCEvaluatedAug 17, 2026
Rank03ModelGOGemini 3 FlashGoogleScore91.8%Percentile95.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank04ModelGOGemini 3 ProGoogleScore91.8%Percentile93.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank05ModelANClaude Opus 4.7AnthropicScore91.5%Percentile91.7%Participants49EvidenceCEvaluatedAug 17, 2026
Rank06ModelANClaude Opus 4.6AnthropicScore91.1%Percentile89.6%Participants49EvidenceCEvaluatedAug 17, 2026
Rank07ModelANClaude Opus 4.5AnthropicScore90.8%Percentile87.5%Participants49EvidenceCEvaluatedAug 17, 2026
Rank08ModelACQwen3.7 MaxAlibaba Cloud / Qwen TeamScore90.3%Percentile85.4%Participants49EvidenceCEvaluatedAug 17, 2026
Rank09ModelOPGPT-5.2OpenAIScore89.6%Percentile83.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank10ModelANClaude Opus 4.1AnthropicScore89.5%Percentile81.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank11ModelACQwen3.6 PlusAlibaba Cloud / Qwen TeamScore89.5%Percentile79.2%Participants49EvidenceCEvaluatedAug 17, 2026
Rank12ModelANClaude Sonnet 4.6AnthropicScore89.3%Percentile77.1%Participants49EvidenceCEvaluatedAug 17, 2026
Rank13ModelANClaude Sonnet 4.5AnthropicScore89.1%Percentile75.0%Participants49EvidenceCEvaluatedAug 17, 2026
Rank14ModelACQwen3.7-PlusAlibaba Cloud / Qwen TeamScore89.0%Percentile72.9%Participants49EvidenceCEvaluatedAug 17, 2026
Rank15ModelGOGemini 3.1 Flash-LiteGoogleScore88.9%Percentile70.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank16ModelANClaude Opus 4AnthropicScore88.8%Percentile68.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank17ModelACQwen3.5-397B-A17BAlibaba Cloud / Qwen TeamScore88.5%Percentile66.7%Participants49EvidenceCEvaluatedAug 17, 2026
Rank18ModelGOGemma 4 31BGoogleScore88.4%Percentile64.6%Participants49EvidenceCEvaluatedAug 17, 2026
Rank19ModelOPo1OpenAIScore87.7%Percentile62.5%Participants49EvidenceCEvaluatedAug 17, 2026
Rank20ModelOPGPT-4.1OpenAIScore87.3%Percentile60.4%Participants49EvidenceCEvaluatedAug 17, 2026
Rank21ModelACQwen3 235B A22BAlibaba Cloud / Qwen TeamScore86.7%Percentile58.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank22ModelACQwen3.5-122B-A10BAlibaba Cloud / Qwen TeamScore86.7%Percentile56.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank23ModelANClaude Sonnet 4AnthropicScore86.5%Percentile54.2%Participants49EvidenceCEvaluatedAug 17, 2026
Rank24ModelGOGemma 4 26B-A4BGoogleScore86.3%Percentile52.1%Participants49EvidenceCEvaluatedAug 17, 2026
Rank25ModelANClaude 3.7 SonnetAnthropicScore86.1%Percentile50.0%Participants49EvidenceCEvaluatedAug 17, 2026
Rank26ModelACQwen3.5-27BAlibaba Cloud / Qwen TeamScore85.9%Percentile47.9%Participants49EvidenceCEvaluatedAug 17, 2026
Rank27ModelLAK-EXAONE-236B-A23BLG AI ResearchScore85.7%Percentile45.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank28ModelMAMistral Large 3 (675B Instruct 2512 Eagle)Mistral AIScore85.5%Percentile43.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank29ModelMAMistral Large 3 (675B Base)Mistral AIScore85.5%Percentile41.7%Participants49EvidenceCEvaluatedAug 17, 2026
Rank30ModelMAMistral Large 3 (675B Instruct 2512 NVFP4)Mistral AIScore85.5%Percentile39.6%Participants49EvidenceCEvaluatedAug 17, 2026
Rank31ModelMAMistral Large 3 (675B Instruct 2512)Mistral AIScore85.5%Percentile37.5%Participants49EvidenceCEvaluatedAug 17, 2026
Rank32ModelACQwen3.5-35B-A3BAlibaba Cloud / Qwen TeamScore85.2%Percentile35.4%Participants49EvidenceCEvaluatedAug 17, 2026
Rank33ModelOPGPT-4.5OpenAIScore85.1%Percentile33.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank34ModelOPGPT OSS 120B HighOpenAIScore83.8%Percentile31.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank35ModelGOGemma 4 12BGoogleScore83.4%Percentile29.2%Participants49EvidenceCEvaluatedAug 17, 2026
Rank36ModelANClaude Haiku 4.5AnthropicScore83.0%Percentile27.1%Participants49EvidenceCEvaluatedAug 17, 2026
Rank37ModelGODiffusionGemma 26B-A4BGoogleScore81.5%Percentile25.0%Participants49EvidenceCEvaluatedAug 17, 2026
Rank38ModelOPGPT-4oOpenAIScore81.4%Percentile22.9%Participants49EvidenceCEvaluatedAug 17, 2026
Rank39ModelACQwen3.5-9BAlibaba Cloud / Qwen TeamScore81.2%Percentile20.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank40ModelOPGPT-4.1 miniOpenAIScore78.5%Percentile18.8%Participants49EvidenceCEvaluatedAug 17, 2026
Rank41ModelGOGemma 4 E4BGoogleScore76.6%Percentile16.7%Participants49EvidenceCEvaluatedAug 17, 2026
Rank42ModelACQwen3.5-4BAlibaba Cloud / Qwen TeamScore76.1%Percentile14.6%Participants49EvidenceCEvaluatedAug 17, 2026
Rank43ModelMAMistral Large 3Mistral AIScore74.2%Percentile12.5%Participants49EvidenceCEvaluatedAug 17, 2026
Rank44ModelMIPhi-3.5-MoE-instructMicrosoftScore69.9%Percentile10.4%Participants49EvidenceCEvaluatedAug 17, 2026
Rank45ModelGOGemma 4 E2BGoogleScore67.4%Percentile8.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank46ModelOPGPT-4.1 nanoOpenAIScore66.9%Percentile6.3%Participants49EvidenceCEvaluatedAug 17, 2026
Rank47ModelACQwen3.5-2BAlibaba Cloud / Qwen TeamScore63.1%Percentile4.2%Participants49EvidenceCEvaluatedAug 17, 2026
Rank48ModelMIPhi-3.5-mini-instructMicrosoftScore55.4%Percentile2.1%Participants49EvidenceCEvaluatedAug 17, 2026
Rank49ModelACQwen3.5-0.8BAlibaba Cloud / Qwen TeamScore44.3%Percentile0.0%Participants49EvidenceCEvaluatedAug 17, 2026

MMMLU Highlights

The leading models and scores on this benchmark.

Rank #1Claude Mythos Preview92.7%Rank #2Gemini 3.1 Pro92.6%Rank #3Gemini 3 Flash91.8%Rank #4Gemini 3 Pro91.8%

MMMLU Score Distribution

A closer view of the leading scores on this benchmark.

MMMLU

The Top AI Models for MMMLU

The first five results on this benchmark, with official price and output speed added where the model identity can be matched.

Ranking basisThis mmmlu AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.

  1. 01
    AN
    Claude Mythos PreviewAnthropic
    Score
    92.7%

    Strengths

    • Ranks #1 of 49 compared models
    • 100th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMLU, not total model capability
  2. 02
    GO
    Gemini 3.1 ProGoogle
    Score
    92.6%
    Price
    $2.0 input / $12 output per 1M tokens
    Speed
    Up to 23 tok/s via Google

    Strengths

    • Ranks #2 of 49 compared models
    • 98th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMLU, not total model capability
  3. 03
    GO
    Gemini 3 FlashGoogle
    Score
    91.8%
    Price
    $0.50 input / $3.0 output per 1M tokens
    Speed
    Up to 65 tok/s via Google

    Strengths

    • Ranks #3 of 49 compared models
    • 96th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMLU, not total model capability
  4. 04
    GO
    Gemini 3 ProGoogle
    Score
    91.8%
    Speed
    Up to 90 tok/s via Google

    Strengths

    • Ranks #4 of 49 compared models
    • 94th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMLU, not total model capability
  5. 05
    AN
    Claude Opus 4.7Anthropic
    Score
    91.5%
    Price
    $5.0 input / $25 output per 1M tokens
    Speed
    Up to 42 tok/s via Anthropic

    Strengths

    • Ranks #5 of 49 compared models
    • 92th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMLU, not total model capability

Selection summary

Best AI Models for MMMLU

Claude Mythos Preview currently leads MMMLU with 92.7%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.

Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.

Benchmark rank #1Claude Mythos Preview92.7%Benchmark rank #2Gemini 3.1 Pro92.6% · $2.0 input / $12 output per 1M tokensBenchmark rank #3Gemini 3 Flash91.8% · $0.50 input / $3.0 output per 1M tokens

What is MMMLU?

What MMMLU measures and how its scores work.

Multilingual Massive Multitask Language Understanding dataset released by OpenAI, featuring professionally translated MMLU test questions across 14 languages including Arabic, Bengali, German, Spanish, French, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Swahili, Yoruba, and Chinese. Contains approximately 15,908 multiple-choice questions per language covering 57 subjects.

Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.

Family
MMMLU
Modality
text
Primary category
math
Score direction
higher
LLMBoard eligible
Yes
Evaluation key
mmmlu|llm-stats-current

Benchmark scores retain their original unit. Overall score eligibility is shown separately.

FAQ

Common questions about MMMLU.

Which model scores highest on MMMLU?

Claude Mythos Preview is currently ranked first with 92.7%.

What does MMMLU measure?

Multilingual Massive Multitask Language Understanding dataset released by OpenAI, featuring professionally translated MMLU test questions across 14 languages including Arabic, Bengali, German, Spanish, French, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Swahili, Yoruba, and Chinese. Contains approximately 15,908 multiple-choice questions per language covering 57 subjects.

Is a higher score better?

Yes. Higher values rank better for this benchmark.

How many models are compared?

49 model results are currently shown.

Does this benchmark affect the overall score?

Yes. This benchmark can contribute to the current LLMBoard capability score.