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

multimodal benchmark

MMMU Leaderboard

MMMU (Massive Multi-discipline Multimodal Understanding) is a benchmark designed to evaluate multimodal models on college-level subject knowledge and deliberate reasoning. Contains 11.5K meticulously collected multimodal questions from college exams, quizzes, and textbooks, covering six core disciplines: Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering across 30 subjects and 183 subfields.

Updated Aug 17, 2026

Models63
Model coverage63
MetricScore
EvidenceB

On this page

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

MMMU Ranking

Higher score ranks better on this benchmark.

30 of 63 rows
Columns

Show columns

Sort by
Rank
Model
Score
Percentile
Participants
Evidence
Evaluated
Rank01ModelACQwen3.6 PlusAlibaba Cloud / Qwen TeamScore86.0%Percentile100.0%Participants63EvidenceCEvaluatedAug 17, 2026
Rank02ModelOPGPT-5.1OpenAIScore85.4%Percentile98.4%Participants63EvidenceCEvaluatedAug 17, 2026
Rank03ModelOPGPT-5.1 InstantOpenAIScore85.4%Percentile96.8%Participants63EvidenceCEvaluatedAug 17, 2026
Rank04ModelOPGPT-5.1 ThinkingOpenAIScore85.4%Percentile95.2%Participants63EvidenceCEvaluatedAug 17, 2026
Rank05ModelOPGPT-5OpenAIScore84.2%Percentile93.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank06ModelACQwen3.5-122B-A10BAlibaba Cloud / Qwen TeamScore83.9%Percentile91.9%Participants63EvidenceCEvaluatedAug 17, 2026
Rank07ModelOPo3OpenAIScore82.9%Percentile90.3%Participants63EvidenceCEvaluatedAug 17, 2026
Rank08ModelACQwen3.6-27BAlibaba Cloud / Qwen TeamScore82.9%Percentile88.7%Participants63EvidenceCEvaluatedAug 17, 2026
Rank09ModelACQwen3.5-27BAlibaba Cloud / Qwen TeamScore82.3%Percentile87.1%Participants63EvidenceCEvaluatedAug 17, 2026
Rank10ModelGOGemini 2.5 Pro Preview 06-05GoogleScore82.0%Percentile85.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank11ModelACQwen3.6-35B-A3BAlibaba Cloud / Qwen TeamScore81.7%Percentile83.9%Participants63EvidenceCEvaluatedAug 17, 2026
Rank12ModelOPo4-miniOpenAIScore81.6%Percentile82.3%Participants63EvidenceCEvaluatedAug 17, 2026
Rank13ModelACQwen3.5-35B-A3BAlibaba Cloud / Qwen TeamScore81.4%Percentile80.7%Participants63EvidenceCEvaluatedAug 17, 2026
Rank14ModelGOGemini 2.5 FlashGoogleScore79.7%Percentile79.0%Participants63EvidenceCEvaluatedAug 17, 2026
Rank15ModelGOGemini 2.5 ProGoogleScore79.6%Percentile77.4%Participants63EvidenceCEvaluatedAug 17, 2026
Rank16ModelSTStep3-VL-10BStepFunScore78.1%Percentile75.8%Participants63EvidenceCEvaluatedAug 17, 2026
Rank17ModelXAGrok-3xAIScore78.0%Percentile74.2%Participants63EvidenceCEvaluatedAug 17, 2026
Rank18ModelOPo1OpenAIScore77.6%Percentile72.6%Participants63EvidenceCEvaluatedAug 17, 2026
Rank19ModelGOGemini 2.0 Flash ThinkingGoogleScore75.4%Percentile71.0%Participants63EvidenceCEvaluatedAug 17, 2026
Rank20ModelOPGPT-4.5OpenAIScore75.2%Percentile69.3%Participants63EvidenceCEvaluatedAug 17, 2026
Rank21ModelCOCommand A+CohereScore75.1%Percentile67.7%Participants63EvidenceCEvaluatedAug 17, 2026
Rank22ModelANClaude 3.7 SonnetAnthropicScore75.0%Percentile66.1%Participants63EvidenceCEvaluatedAug 17, 2026
Rank23ModelOPGPT-4.1OpenAIScore74.8%Percentile64.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank24ModelANClaude Sonnet 4AnthropicScore74.4%Percentile62.9%Participants63EvidenceCEvaluatedAug 17, 2026
Rank25ModelMELlama 4 MaverickMetaScore73.4%Percentile61.3%Participants63EvidenceCEvaluatedAug 17, 2026
Rank26ModelGOGemini 2.5 Flash-LiteGoogleScore72.9%Percentile59.7%Participants63EvidenceCEvaluatedAug 17, 2026
Rank27ModelOPGPT-4.1 miniOpenAIScore72.7%Percentile58.1%Participants63EvidenceCEvaluatedAug 17, 2026
Rank28ModelOPGPT-4oOpenAIScore72.2%Percentile56.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank29ModelGOGemini 2.0 FlashGoogleScore70.7%Percentile54.8%Participants63EvidenceCEvaluatedAug 17, 2026
Rank30ModelACQvQ-72B-PreviewAlibaba Cloud / Qwen TeamScore70.3%Percentile53.2%Participants63EvidenceCEvaluatedAug 17, 2026
Rank31ModelACQwen2.5 VL 72B InstructAlibaba Cloud / Qwen TeamScore70.2%Percentile51.6%Participants63EvidenceCEvaluatedAug 17, 2026
Rank32ModelMAKimi-k1.5Moonshot AIScore70.0%Percentile50.0%Participants63EvidenceCEvaluatedAug 17, 2026
Rank33ModelACQwen2.5 VL 32B InstructAlibaba Cloud / Qwen TeamScore70.0%Percentile48.4%Participants63EvidenceCEvaluatedAug 17, 2026
Rank34ModelMELlama 4 ScoutMetaScore69.4%Percentile46.8%Participants63EvidenceCEvaluatedAug 17, 2026
Rank35ModelANClaude 3.5 SonnetAnthropicScore68.3%Percentile45.2%Participants63EvidenceCEvaluatedAug 17, 2026
Rank36ModelGOGemini 2.0 Flash-LiteGoogleScore68.0%Percentile43.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank37ModelXAGrok-2xAIScore66.1%Percentile41.9%Participants63EvidenceCEvaluatedAug 17, 2026
Rank38ModelGOGemini 1.5 ProGoogleScore65.9%Percentile40.3%Participants63EvidenceCEvaluatedAug 17, 2026
Rank39ModelMAPixtral LargeMistral AIScore64.0%Percentile38.7%Participants63EvidenceCEvaluatedAug 17, 2026
Rank40ModelXAGrok-2 minixAIScore63.2%Percentile37.1%Participants63EvidenceCEvaluatedAug 17, 2026
Rank41ModelMAMistral Small 3.2 24B InstructMistral AIScore62.5%Percentile35.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank42ModelGOGemini 1.5 FlashGoogleScore62.3%Percentile33.9%Participants63EvidenceCEvaluatedAug 17, 2026
Rank43ModelAMNova ProAmazonScore61.7%Percentile32.3%Participants63EvidenceCEvaluatedAug 17, 2026
Rank44ModelMELlama 3.2 90B InstructMetaScore60.3%Percentile30.6%Participants63EvidenceCEvaluatedAug 17, 2026
Rank45ModelOPGPT-4o miniOpenAIScore59.4%Percentile29.0%Participants63EvidenceCEvaluatedAug 17, 2026
Rank46ModelMAMistral Small 3.1 24B BaseMistral AIScore59.3%Percentile27.4%Participants63EvidenceCEvaluatedAug 17, 2026
Rank47ModelMAMistral Small 3.1 24B InstructMistral AIScore59.3%Percentile25.8%Participants63EvidenceCEvaluatedAug 17, 2026
Rank48ModelACQwen2.5-Omni-7BAlibaba Cloud / Qwen TeamScore59.2%Percentile24.2%Participants63EvidenceCEvaluatedAug 17, 2026
Rank49ModelACQwen2.5 VL 7B InstructAlibaba Cloud / Qwen TeamScore58.6%Percentile22.6%Participants63EvidenceCEvaluatedAug 17, 2026
Rank50ModelAMNova LiteAmazonScore56.2%Percentile21.0%Participants63EvidenceCEvaluatedAug 17, 2026
Rank51ModelOPGPT-4.1 nanoOpenAIScore55.4%Percentile19.4%Participants63EvidenceCEvaluatedAug 17, 2026
Rank52ModelMIPhi-4-multimodal-instructMicrosoftScore55.1%Percentile17.7%Participants63EvidenceCEvaluatedAug 17, 2026
Rank53ModelGOGemini 1.5 Flash 8BGoogleScore53.7%Percentile16.1%Participants63EvidenceCEvaluatedAug 17, 2026
Rank54ModelXAGrok-1.5xAIScore53.6%Percentile14.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank55ModelXAGrok-1.5VxAIScore53.6%Percentile12.9%Participants63EvidenceCEvaluatedAug 17, 2026
Rank56ModelMAPixtral-12BMistral AIScore52.5%Percentile11.3%Participants63EvidenceCEvaluatedAug 17, 2026
Rank57ModelDEDeepSeek VL2DeepSeekScore51.1%Percentile9.7%Participants63EvidenceCEvaluatedAug 17, 2026
Rank58ModelMELlama 3.2 11B InstructMetaScore50.7%Percentile8.1%Participants63EvidenceCEvaluatedAug 17, 2026
Rank59ModelDEDeepSeek VL2 SmallDeepSeekScore48.0%Percentile6.5%Participants63EvidenceCEvaluatedAug 17, 2026
Rank60ModelGOGemini 1.0 ProGoogleScore47.9%Percentile4.8%Participants63EvidenceBEvaluatedAug 17, 2026
Rank61ModelMIPhi-3.5-vision-instructMicrosoftScore43.0%Percentile3.2%Participants63EvidenceCEvaluatedAug 17, 2026
Rank62ModelDEDeepSeek VL2 TinyDeepSeekScore40.7%Percentile1.6%Participants63EvidenceCEvaluatedAug 17, 2026
Rank63ModelOPGPT-3.5 TurboOpenAIScore0.0%Percentile0.0%Participants63EvidenceBEvaluatedAug 17, 2026

MMMU Highlights

The leading models and scores on this benchmark.

Rank #1Qwen3.6 Plus86.0%Rank #2GPT-5.185.4%Rank #3GPT-5.1 Instant85.4%Rank #4GPT-5.1 Thinking85.4%

MMMU Score Distribution

A closer view of the leading scores on this benchmark.

MMMU

The Top AI Models for MMMU

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

Ranking basisThis mmmu 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
    AC
    Qwen3.6 PlusAlibaba Cloud / Qwen Team
    Score
    86.0%
    Price
    $0.50 input / $3.0 output per 1M tokens
    Speed
    Up to 16 tok/s via Together

    Strengths

    • Ranks #1 of 63 compared models
    • 100th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMU, not total model capability
  2. 02
    OP
    GPT-5.1OpenAI
    Score
    85.4%
    Price
    $1.3 input / $10 output per 1M tokens
    Speed
    Up to 100 tok/s via OpenAI

    Strengths

    • Ranks #2 of 63 compared models
    • 98th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMU, not total model capability
  3. 03
    OP
    GPT-5.1 InstantOpenAI
    Score
    85.4%
    Speed
    Up to 100 tok/s via OpenAI

    Strengths

    • Ranks #3 of 63 compared models
    • 97th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMU, not total model capability
  4. 04
    OP
    GPT-5.1 ThinkingOpenAI
    Score
    85.4%
    Price
    $1.3 input / $10 output per 1M tokens
    Speed
    Up to 80 tok/s via OpenAI

    Strengths

    • Ranks #4 of 63 compared models
    • 95th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMU, not total model capability
  5. 05
    OP
    GPT-5OpenAI
    Score
    84.2%
    Price
    $1.3 input / $10 output per 1M tokens
    Speed
    Up to 100 tok/s via OpenAI

    Strengths

    • Ranks #5 of 63 compared models
    • 94th percentile on this benchmark
    • C evidence result

    Considerations

    • This result measures MMMU, not total model capability

Selection summary

Best AI Models for MMMU

Qwen3.6 Plus currently leads MMMU with 86.0%. 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 #1Qwen3.6 Plus86.0% · $0.50 input / $3.0 output per 1M tokensBenchmark rank #2GPT-5.185.4% · $1.3 input / $10 output per 1M tokensBenchmark rank #3GPT-5.1 Instant85.4% · Up to 100 tok/s via OpenAI

What is MMMU?

What MMMU measures and how its scores work.

MMMU (Massive Multi-discipline Multimodal Understanding) is a benchmark designed to evaluate multimodal models on college-level subject knowledge and deliberate reasoning. Contains 11.5K meticulously collected multimodal questions from college exams, quizzes, and textbooks, covering six core disciplines: Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering across 30 subjects and 183 subfields.

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

Family
MMMU
Modality
multimodal
Primary category
multimodal
Score direction
higher
LLMBoard eligible
Yes
Evaluation key
mmmu|llm-stats-current

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

FAQ

Common questions about MMMU.

Which model scores highest on MMMU?

Qwen3.6 Plus is currently ranked first with 86.0%.

What does MMMU measure?

MMMU (Massive Multi-discipline Multimodal Understanding) is a benchmark designed to evaluate multimodal models on college-level subject knowledge and deliberate reasoning. Contains 11.5K meticulously collected multimodal questions from college exams, quizzes, and textbooks, covering six core disciplines: Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering across 30 subjects and 183 subfields.

Is a higher score better?

Yes. Higher values rank better for this benchmark.

How many models are compared?

63 model results are currently shown.

Does this benchmark affect the overall score?

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