MMMU
The official leaderboard hasn’t scored a model released after Reka Edge (March 2026). Labs report their own numbers for 10 newer models, such as Qwen 3.6 Plus at 86%.
11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy.
Near-saturated. The test set answers became public on February 12th, 2026, so test scores can now be computed locally.
- Released
- November 2023
- Kind
- General visual questions
- Level
- University
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- All public
- Measures
- Capability
- Subject
- Arts, Economics and business, Other natural sciences, Clinical medicine, Other engineering and technologies
- Language
- English
- Interaction
- Single turn
- Refresh
- New versions replace it
- Run by
- Anyone
- In views
- Vision, Biology and medicine, Law and finance
- Strongest model tested
- Claude Opus 4.5 #62 on the capability index
- Last in a launch post
- October 2026
- Last model added
- October 2026
- Scores refreshed
- November 18th, 2026
In the authors’ words
We introduce MMMU: a new benchmark designed to evaluate multimodal models on massive multi-discipline tasks demanding college-level subject knowledge and deliberate reasoning. MMMU includes 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. These questions span 30 subjects and 183 subfields, comprising 30 highly heterogeneous image types, such as charts, diagrams, maps, tables, music sheets, and chemical structures. Unlike existing benchmarks, MMMU focuses on advanced perception and reasoning with domain-specific knowledge, challenging models to perform tasks akin to those faced by experts. The evaluation of 14 open-source LMMs as well as the proprietary GPT-4V(ision) and Gemini highlights the substantial challenges posed by MMMU. Even the advanced GPT-4V and Gemini Ultra only achieve accuracies of 56% and 59% respectively, indicating significant room for improvement. We believe MMMU will stimulate the community to build next-generation multimodal foundation models towards expert artificial general intelligence.
Qwen 3.6 Plus leads with 86%
Best score per model as labs report them in their own cards, drawn striped. Labs pick their own settings, so these compare less cleanly than independent runs.
GPT-5.1 leads with 85.4% across all independent runs
Best score per model across the official leaderboard, Epoch AI’s own runs and any board Epoch AI copied. Dotted bars are reruns by Epoch AI or another board, and each row names its source.
Qwen 3.6 Plus leads with 86%, claims included
Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.
- 86.0
Qwen 3.6 Plus
- 85.4 GPT-5.1
- 85.4
Seed2.0 Pro
- 85.0
Qwen3.5 397B-A17B
- 84.3 Kimi K2.5
- 84.2 GPT-5
- 83.9
Qwen3.5-122B-A10B
- 83.4
Seed1.8
- 82.9 o3
- 82.9
Qwen3.6 27B
- 82.3
Qwen3.5-27B
- 82.0
Gemini 2.5 Pro
- 81.7
Gemini 2.5 Pro
- 81.7
Qwen 3.6 35B-A3B
- 81.6 o4-mini
- 81.6 GPT-5 mini
- 81.4
Qwen3.5-35B-A3B
- 80.7
gemini-robotics-er-1.5-preview
- 80.7 Claude Opus 4.5
- 80.6
Qwen3-VL-235B-A22B
- 80.3
Gemini 2.5 Flash
- 80.2
MiniCPM-V-4.7-35B-A3B
- 80.1
Qwen3.5-Omni-Plus
- 79.7
Gemini 2.5 Flash
- 78.7
EXAONE 4.5 33B
- 78.4
Qwen3.5-9B
- 78.1
STEP3-VL-10B
- 78.0 Grok 3
- 77.9
Seed1.5-VL
- 77.8 Claude Sonnet 4.5
- 77.7
Intern-S1
- 77.7
InternVL3.5-241B-A28B
- 77.6
Qwen3.5-4B
- 77.3 o1
- 77.1 Claude Opus 4.1
- 76.7
Gemini 2.5 Flash
- 76.5 Claude Opus 4
- 76.1
Llama 4 Behemoth
- 76.0 GLM-4.6V
- 75.6 GPT-5 nano
- 75.4
Gemini 2.0 Flash Thinking
- 75.4 GLM-4.5V
- 75.3
Keye-VL-2.0-30B-A3B
- 75.1
Command A+
- 75.0 Claude 3.7 Sonnet
- 74.8 GPT-4.1
- 74.8
Seed1.6-Thinking
- 74.4 GPT-4.5
- 74.4 Claude Sonnet 4
- 74.2
Step 3
- 73.4
Llama 4 Maverick
- 73.2 Claude Haiku 4.5
- 72.9
Gemini 2.5 Flash-Lite
- 72.7
Gemini 2.0 Pro
- 72.7 GPT-4.1 mini
- 72.2
ERNIE-4.5-VL-28B-A3B-Thinking
- 72.0
Gemini 2.5 Flash-Lite
- 71.7
Gemini 2.0 Flash
- 70.8
Nemotron 3 Nano Omni 30B-A3B
- 70.7
Gemini 2.0 Flash
- 70.7
LongCat-Flash-Omni
- 70.4 Claude 3.5 Sonnet (October 2024)
- 70.2
Qwen2.5-VL-72B
- 70.0 Kimi k1.5
- 70.0
Magistral Medium 1.0
- 69.4 Grok-3 mini
- 69.4
Llama 4 Scout
- 69.1 GPT-4o
- 69.1
Qwen3-Omni-30B-A3B
- 68.3 Claude 3.5 Sonnet
- 68.0
Gemini 2.0 Flash-Lite
- 68.0
Amazon Nova Premier
- 68.0 GLM-4.1V-9B-Thinking
- 68.0
Nemotron Nano 12B v2 VL
- 66.1 Grok-2
- 66.1
Mistral Medium 3
- 66.0
Magistral Small 1.0
- 64.9
Gemma 3 27B
- 64.2
Qwen3.5-2B
- 64.0
Mistral Small 3.1
- 64.0 Kimi-VL-A3B-Thinking-2506
- 62.5
Mistral Small 3.2
- 62.4
Gemini 1.0 Ultra
- 62.2
Gemini 1.5 Pro
- 61.7 Kimi-VL-A3B-Thinking
- 59.6
Gemma 3 12B
- 59.4 Claude 3 Opus
- 59.4 GPT-4o mini
- 59.2
Qwen2.5-Omni-7B
- 58.5
Gemini 1.5 Pro
- 55.4 GPT-4.1 nano
- 55.3
BAGEL-7B-MoT
- 55.1
Phi-4-multimodal
- 54.3
Phi-4-Reasoning-Vision-15B
- 53.1 Claude 3 Sonnet
- 50.2 Claude 3 Haiku
- 49.0
Qwen3.5-0.8B
- 48.8
Gemma 3 4B
- 48.4 LFM2.5-VL-3B
- 48.2
Llama-3.1-Nemotron-Nano-VL-8B-V1
- 43.0 Phi-3.5-vision-instruct
- 41.0
Janus-Pro-7B
- 39.7 LFM2-VL-1.6B
- 32.9
North Micro Vision Instruct
- 1
Qwen 3.6 Plus 86%
- 2 GPT-5.1 85.4%
- 3
Seed2.0 Pro 85.4%
- 4
Qwen3.5 397B-A17B 85%
- 5 Kimi K2.5 84.3%
- 6 GPT-5 84.2%
- 7
Qwen3.5-122B-A10B 83.9%
- 8
Seed1.8 83.4%
- 9 o3 82.9%
- 10
Qwen3.6 27B 82.9%
- 11
Qwen3.5-27B 82.3%
- 12
Gemini 2.5 Pro Jun 2025 82%
- 13
Gemini 2.5 Pro Mar 2025 81.7%
- 14
Qwen 3.6 35B-A3B 81.7%
- 15 o4-mini 81.6%
- 16 GPT-5 mini 81.6%
- 17
Qwen3.5-35B-A3B 81.4%
- 18
gemini-robotics-er-1.5-preview 80.7%
- 19 Claude Opus 4.5 80.7%
- 20
Qwen3-VL-235B-A22B 80.6%
- 21
Gemini 2.5 Flash Sep 2025 80.3%
- 22
MiniCPM-V-4.7-35B-A3B 80.2%
- 23
Qwen3.5-Omni-Plus 80.1%
- 24
Gemini 2.5 Flash Jun 2025 79.7%
- 25
EXAONE 4.5 33B 78.7%
- 26
Qwen3.5-9B 78.4%
- 27
STEP3-VL-10B 78.1%
- 28 Grok 3 78%
- 29
Seed1.5-VL 77.9%
- 30 Claude Sonnet 4.5 77.8%
- 31
Intern-S1 77.7%
- 32
InternVL3.5-241B-A28B 77.7%
- 33
Qwen3.5-4B 77.6%
- 34 o1 77.3%
- 35 Claude Opus 4.1 77.1%
- 36
Gemini 2.5 Flash Apr 2025 76.7%
- 37 Claude Opus 4 76.5%
- 38
Llama 4 Behemoth 76.1%
- 39 GLM-4.6V 76%
- 40 GPT-5 nano 75.6%
- 41
Gemini 2.0 Flash Thinking Jan 2025 75.4%
- 42 GLM-4.5V 75.4%
- 43
Keye-VL-2.0-30B-A3B 75.3%
- 44
Command A+ 75.1%
- 45 Claude 3.7 Sonnet 75%
- 46 GPT-4.1 74.8%
- 47
Seed1.6-Thinking 74.8%
- 48 GPT-4.5 74.4%
- 49 Claude Sonnet 4 74.4%
- 50
Step 3 74.2%
- 51
Llama 4 Maverick 73.4%
- 52 Claude Haiku 4.5 73.2%
- 53
Gemini 2.5 Flash-Lite Jun 2025 72.9%
- 54
Gemini 2.0 Pro 72.7%
- 55 GPT-4.1 mini 72.7%
- 56
ERNIE-4.5-VL-28B-A3B-Thinking 72.2%
- 57
Gemini 2.5 Flash-Lite Sep 2025 72%
- 58
Gemini 2.0 Flash Feb 2025 71.7%
- 59
Nemotron 3 Nano Omni 30B-A3B 70.8%
- 60
Gemini 2.0 Flash Dec 2024 70.7%
- 61
LongCat-Flash-Omni 70.7%
- 62 Claude 3.5 Sonnet (October 2024) 70.4%
- 63
Qwen2.5-VL-72B 70.2%
- 64 Kimi k1.5 70%
- 65
Magistral Medium 1.0 70%
- 66 Grok-3 mini 69.4%
- 67
Llama 4 Scout 69.4%
- 68 GPT-4o May 2024 69.1%
- 69
Qwen3-Omni-30B-A3B 69.1%
- 70 Claude 3.5 Sonnet 68.3%
- 71
Gemini 2.0 Flash-Lite 68%
- 72
Amazon Nova Premier 68%
- 73 GLM-4.1V-9B-Thinking 68%
- 74
Nemotron Nano 12B v2 VL 68%
- 75 Grok-2 Aug 2024 66.1%
- 76
Mistral Medium 3 66.1%
- 77
Magistral Small 1.0 66%
- 78
Gemma 3 27B 64.9%
- 79
Qwen3.5-2B 64.2%
- 80
Mistral Small 3.1 64%
- 81 Kimi-VL-A3B-Thinking-2506 64%
- 82
Mistral Small 3.2 62.5%
- 83
Gemini 1.0 Ultra 62.4%
- 84
Gemini 1.5 Pro May 2024 62.2%
- 85 Kimi-VL-A3B-Thinking 61.7%
- 86
Gemma 3 12B 59.6%
- 87 Claude 3 Opus 59.4%
- 88 GPT-4o mini 59.4%
- 89
Qwen2.5-Omni-7B 59.2%
- 90
Gemini 1.5 Pro Feb 2024 58.5%
- 91 GPT-4.1 nano 55.4%
- 92
BAGEL-7B-MoT 55.3%
- 93
Phi-4-multimodal 55.1%
- 94
Phi-4-Reasoning-Vision-15B 54.3%
- 95 Claude 3 Sonnet 53.1%
- 96 Claude 3 Haiku 50.2%
- 97
Qwen3.5-0.8B 49%
- 98
Gemma 3 4B 48.8%
- 99 LFM2.5-VL-3B 48.4%
- 100
Llama-3.1-Nemotron-Nano-VL-8B-V1 48.2%
- 101 Phi-3.5-vision-instruct 43%
- 102
Janus-Pro-7B 41%
- 103 LFM2-VL-1.6B 39.7%
- 104
North Micro Vision Instruct 32.9%
Source: Labs’ model cards
- 85.4 GPT-5.1
- 84.2 GPT-5
- 84.0
Gemini 2.5 Deep Think
- 82.9 o3
- 81.6 o4-mini
- 80.7 Claude Opus 4.5
- 80.1 dots.vlm1
- 79.7
Gemini 2.5 Flash
- 79.6
Gemini 2.5 Pro
- 78.7
Qwen3-VL-235B-A22B
- 78.2 o1
- 78.0 Grok 3
- 77.9
Seed1.5-VL
- 77.8 Claude Sonnet 4.5
- 77.7
Intern-S1
- 77.1 Claude Opus 4.1
- 76.5 Claude Opus 4
- 76.1
Llama 4 Behemoth
- 76.0
Skywork-R1V3-38B
- 75.4 GLM-4.5V
- 75.0 Claude 3.7 Sonnet
- 74.8
Seed1.6-Thinking
- 74.4 GPT-4.5
- 74.4 Claude Sonnet 4
- 73.6
Skywork-R1V2-38B
- 73.4
Llama 4 Maverick
- 72.7
Gemini 2.0 Pro
- 72.2
InternVL3-78B
- 71.7
Gemini 2.0 Flash
- 70.7 GPT-4o
- 70.6 MiMo-VL-7B-RL-2508
- 70.3
QVQ-72B-Preview
- 70.1
InternVL2_5-78B
- 70.1
InternVL3-38B
- 69.4
Llama 4 Scout
- 69.1 GPT-4o
- 69.0
Skywork-R1V-38B
- 68.3 Claude 3.5 Sonnet
- 68.0
Gemini 2.0 Flash-Lite
- 68.0
Amazon Nova Premier
- 68.0 GLM-4.1V-9B-Thinking
- 67.1
InternVL3-14B
- 66.7 MiMo-VL-7B-RL
- 65.6
Ovis2-34B
- 64.9
Gemma 3 27B
- 64.6 MiMo-VL-7B-SFT
- 64.5 Qwen2-VL-72B-Instruct
- 64.5
Qwen2.5-VL-72B
- 63.9
InternVL2.5-38B
- 62.7
InternVL3-8B
- 62.2
Gemini 1.5 Pro
- 62.0
InternVL2-Pro
- 62.0
Amazon Nova Pro
- 60.3
Llama 3.2 90B
- 60.0
InternVL2.5-26B
- 59.7
NVLM-D-72B
- 59.6
Ovis2-16B
- 59.6
Gemma 3 12B
- 59.4
Gemini 1.0 Ultra
- 59.4 Claude 3 Opus
- 59.4 GPT-4o mini
- 59.0
Ovis2-8B
- 58.6
Qwen2.5-VL-7B
- 58.3
InternVL2-Llama3-76B
- 58.3
Qianfan-VL-70B
- 57.7
InternVL3-9B
- 57.0 Kimi-VL-A3B-Instruct
- 56.8 GPT-4V
- 56.8 LLaVA-OneVision 72B
- 56.3 Reka Core
- 56.1
Gemini 1.5 Flash
- 56.0
InternVL2.5-8B
- 55.8
Eagle2.5-8B
- 55.2
InternVL2-40B
- 54.9 Aria
- 53.3 Reka Flash
- 53.1 Claude 3 Sonnet
- 53.1
Qwen2.5-VL-3B
- 51.6
InternVL-Chat-V1-2
- 51.4 Qwen-VL-Max
- 51.2
InternVL2-8B
- 51.1
DeepSeek-VL2
- 50.8
MAmmoTH-VL-8B
- 50.7
Llama 3.2 11B
- 50.2 Claude 3 Haiku
- 49.9
LLaVA-NeXT-72B
- 49.8
Ovis1.5-Gemma2-9B
- 49.8 MiniCPM-V-2_6
- 49.7
Cambrian-1-34B
- 48.8
LLaVA-OneVision 7B
- 48.8
Gemma 3 4B
- 48.6
InternVL3-2B
- 48.1
LLaVA-v1.6-34B
- 48.0
DeepSeek-VL2-Small
- 48.0
Ovis2-4B
- 47.9
Gemini 1.0 Pro
- 47.4 Aquila-VL-2B
- 47.1
Qianfan-VL-8B
- 46.6
Idefics3-8B-Llama3
- 46.4
Qianfan-VL-3B
- 45.9
Yi-VL-34B
- 45.8
MiniCPM-Llama3-V 2.5
- 45.3
Ovis2-2B
- 45.2
Qwen-VL-Plus
- 45.1
Aya Vision 32B
- 43.6
InternVL2.5-2B
- 43.4
InternVL3-1B
- 43.0
InternLM-XComposer2-VL-7B
- 43.0 Phi-3.5-vision-instruct
- 42.8 Reka Edge
- 42.7
Cambrian-1-8B
- 41.4 Bunny-v1.0-4B
- 41.1
OmniLMM-12B
- 40.9
InternVL2.5-1B
- 40.7
DeepSeek-VL2-Tiny
- 40.0
Cambrian-1-13B
- 39.9
Aya Vision 8B
- 39.1
Yi-VL-6B
- 39.1
InternVL-Chat-V1-1
- 38.2 Bunny-v1.0-3B
- 37.2
MiniCPM-V
- 37.1
MiniCPM-V 2.0
- 36.4
LLaVA-v1.5-13B
- 36.3 Emu2-Chat
- 36.2 llava-v1.6-vicuna-13b
- 36.0
Ovis2-1B
- 35.9 Qwen-VL-Chat
- 35.8 llava-v1.6-vicuna-7b
- 35.7
InstructBLIP FLAN-T5-XXL
- 35.4
BLIP-2 FLAN-T5-XXL
- 35.3 llava-v1.6-mistral-7b
- 34.4
BLIP-2 FLAN-T5-XL
- 32.9
InstructBLIP FLAN-T5-XL
- 32.7
mPLUG-Owl2
- 32.1 CogVLM-17B-Chat
- 27.9 Fuyu-8B
- 24.4
Kosmos-2
- 1 GPT-5.1 85.4%
- 2 GPT-5 thinking 84.2%
- 3
Gemini 2.5 Deep Think 84%
- 4 o3 82.9%
- 5 o4-mini 81.6%
- 6 Claude Opus 4.5 80.7%
- 7 dots.vlm1 80.1%
- 8
Gemini 2.5 Flash May 2025 79.7%
- 9
Gemini 2.5 Pro May 2025 79.6%
- 10
Qwen3-VL-235B-A22B 78.7%
- 11 o1 78.2%
- 12 Grok 3 78%
- 13
Seed1.5-VL thinking 77.9%
- 14 Claude Sonnet 4.5 77.8%
- 15
Intern-S1 77.7%
- 16 Claude Opus 4.1 77.1%
- 17 Claude Opus 4 76.5%
- 18
Llama 4 Behemoth 76.1%
- 19
Skywork-R1V3-38B 76%
- 20 GLM-4.5V thinking 75.4%
- 21 Claude 3.7 Sonnet 75%
- 22
Seed1.6-Thinking 74.8%
- 23 GPT-4.5 74.4%
- 24 Claude Sonnet 4 74.4%
- 25
Skywork-R1V2-38B 73.6%
- 26
Llama 4 Maverick 73.4%
- 27
Gemini 2.0 Pro 72.7%
- 28
InternVL3-78B 72.2%
- 29
Gemini 2.0 Flash Feb 2025 71.7%
- 30 GPT-4o Nov 2024 70.7%
- 31 MiMo-VL-7B-RL-2508 thinking 70.6%
- 32
QVQ-72B-Preview 70.3%
- 33
InternVL2_5-78B 70.1%
- 34
InternVL3-38B 70.1%
- 35
Llama 4 Scout 69.4%
- 36 GPT-4o May 2024 69.1%
- 37
Skywork-R1V-38B 69%
- 38 Claude 3.5 Sonnet 68.3%
- 39
Gemini 2.0 Flash-Lite 68%
- 40
Amazon Nova Premier 68%
- 41 GLM-4.1V-9B-Thinking thinking 68%
- 42
InternVL3-14B 67.1%
- 43 MiMo-VL-7B-RL 66.7%
- 44
Ovis2-34B 65.6%
- 45
Gemma 3 27B 64.9%
- 46 MiMo-VL-7B-SFT 64.6%
- 47 Qwen2-VL-72B-Instruct 64.5%
- 48
Qwen2.5-VL-72B 64.5%
- 49
InternVL2.5-38B 63.9%
- 50
InternVL3-8B 62.7%
- 51
Gemini 1.5 Pro May 2024 62.2%
- 52
InternVL2-Pro 62%
- 53
Amazon Nova Pro 62%
- 54
Llama 3.2 90B 60.3%
- 55
InternVL2.5-26B 60%
- 56
NVLM-D-72B 59.7%
- 57
Ovis2-16B 59.6%
- 58
Gemma 3 12B 59.6%
- 59
Gemini 1.0 Ultra 59.4%
- 60 Claude 3 Opus 59.4%
- 61 GPT-4o mini 59.4%
- 62
Ovis2-8B 59%
- 63
Qwen2.5-VL-7B 58.6%
- 64
InternVL2-Llama3-76B 58.3%
- 65
Qianfan-VL-70B 58.3%
- 66
InternVL3-9B 57.7%
- 67 Kimi-VL-A3B-Instruct 57%
- 68 GPT-4V 56.8%
- 69 LLaVA-OneVision 72B 56.8%
- 70 Reka Core 56.3%
- 71
Gemini 1.5 Flash May 2024 56.1%
- 72
InternVL2.5-8B 56%
- 73
Eagle2.5-8B 55.8%
- 74
InternVL2-40B 55.2%
- 75 Aria 54.9%
- 76 Reka Flash 53.3%
- 77 Claude 3 Sonnet 53.1%
- 78
Qwen2.5-VL-3B 53.1%
- 79
InternVL-Chat-V1-2 51.6%
- 80 Qwen-VL-Max 51.4%
- 81
InternVL2-8B 51.2%
- 82
DeepSeek-VL2 51.1%
- 83
MAmmoTH-VL-8B 50.8%
- 84
Llama 3.2 11B 50.7%
- 85 Claude 3 Haiku 50.2%
- 86
LLaVA-NeXT-72B 49.9%
- 87
Ovis1.5-Gemma2-9B 49.8%
- 88 MiniCPM-V-2_6 49.8%
- 89
Cambrian-1-34B 49.7%
- 90
LLaVA-OneVision 7B 48.8%
- 91
Gemma 3 4B 48.8%
- 92
InternVL3-2B 48.6%
- 93
LLaVA-v1.6-34B 48.1%
- 94
DeepSeek-VL2-Small 48%
- 95
Ovis2-4B 48%
- 96
Gemini 1.0 Pro 47.9%
- 97 Aquila-VL-2B 47.4%
- 98
Qianfan-VL-8B 47.1%
- 99
Idefics3-8B-Llama3 46.6%
- 100
Qianfan-VL-3B 46.4%
- 101
Yi-VL-34B 45.9%
- 102
MiniCPM-Llama3-V 2.5 45.8%
- 103
Ovis2-2B 45.3%
- 104
Qwen-VL-Plus 45.2%
- 105
Aya Vision 32B 45.1%
- 106
InternVL2.5-2B 43.6%
- 107
InternVL3-1B 43.4%
- 108
InternLM-XComposer2-VL-7B 43%
- 109 Phi-3.5-vision-instruct 43%
- 110 Reka Edge 42.8%
- 111
Cambrian-1-8B 42.7%
- 112 Bunny-v1.0-4B 41.4%
- 113
OmniLMM-12B 41.1%
- 114
InternVL2.5-1B 40.9%
- 115
DeepSeek-VL2-Tiny 40.7%
- 116
Cambrian-1-13B 40%
- 117
Aya Vision 8B 39.9%
- 118
Yi-VL-6B 39.1%
- 119
InternVL-Chat-V1-1 39.1%
- 120 Bunny-v1.0-3B 38.2%
- 121
MiniCPM-V 37.2%
- 122
MiniCPM-V 2.0 37.1%
- 123
LLaVA-v1.5-13B 36.4%
- 124 Emu2-Chat 36.3%
- 125 llava-v1.6-vicuna-13b 36.2%
- 126
Ovis2-1B 36%
- 127 Qwen-VL-Chat 35.9%
- 128 llava-v1.6-vicuna-7b 35.8%
- 129
InstructBLIP FLAN-T5-XXL 35.7%
- 130
BLIP-2 FLAN-T5-XXL 35.4%
- 131 llava-v1.6-mistral-7b 35.3%
- 132
BLIP-2 FLAN-T5-XL 34.4%
- 133
InstructBLIP FLAN-T5-XL 32.9%
- 134
mPLUG-Owl2 32.7%
- 135 CogVLM-17B-Chat 32.1%
- 136 Fuyu-8B 27.9%
- 137
Kosmos-2 24.4%
Source: Official MMMU leaderboard
- 86.0
Qwen 3.6 Plus
- 85.4 GPT-5.1
- 85.4
Seed2.0 Pro
- 85.0
Qwen3.5 397B-A17B
- 84.3 Kimi K2.5
- 84.2 GPT-5
- 84.0
Gemini 2.5 Deep Think
- 83.9
Qwen3.5-122B-A10B
- 83.4
Seed1.8
- 82.9 o3
- 82.9
Qwen3.6 27B
- 82.3
Qwen3.5-27B
- 82.0
Gemini 2.5 Pro
- 81.7
Gemini 2.5 Pro
- 81.7
Qwen 3.6 35B-A3B
- 81.6 o4-mini
- 81.6 GPT-5 mini
- 81.4
Qwen3.5-35B-A3B
- 80.7
gemini-robotics-er-1.5-preview
- 80.7 Claude Opus 4.5
- 80.6
Qwen3-VL-235B-A22B
- 80.3
Gemini 2.5 Flash
- 80.2
MiniCPM-V-4.7-35B-A3B
- 80.1 dots.vlm1
- 80.1
Qwen3.5-Omni-Plus
- 79.7
Gemini 2.5 Flash
- 79.7
Gemini 2.5 Flash
- 79.6
Gemini 2.5 Pro
- 78.7
EXAONE 4.5 33B
- 78.4
Qwen3.5-9B
- 78.2 o1
- 78.1
STEP3-VL-10B
- 78.0 Grok 3
- 77.9
Seed1.5-VL
- 77.8 Claude Sonnet 4.5
- 77.7
Intern-S1
- 77.7
InternVL3.5-241B-A28B
- 77.6
Qwen3.5-4B
- 77.1 Claude Opus 4.1
- 76.7
Gemini 2.5 Flash
- 76.5 Claude Opus 4
- 76.1
Llama 4 Behemoth
- 76.0
Skywork-R1V3-38B
- 76.0 GLM-4.6V
- 75.6 GPT-5 nano
- 75.4
Gemini 2.0 Flash Thinking
- 75.4 GLM-4.5V
- 75.3
Keye-VL-2.0-30B-A3B
- 75.1
Command A+
- 75.0 Claude 3.7 Sonnet
- 74.8 GPT-4.1
- 74.8
Seed1.6-Thinking
- 74.4 GPT-4.5
- 74.4 Claude Sonnet 4
- 74.2
Step 3
- 73.6
Skywork-R1V2-38B
- 73.4
Llama 4 Maverick
- 73.2 Claude Haiku 4.5
- 72.9
Gemini 2.5 Flash-Lite
- 72.7
Gemini 2.0 Pro
- 72.7 GPT-4.1 mini
- 72.2
InternVL3-78B
- 72.2
ERNIE-4.5-VL-28B-A3B-Thinking
- 72.0
Gemini 2.5 Flash-Lite
- 71.7
Gemini 2.0 Flash
- 70.8
Nemotron 3 Nano Omni 30B-A3B
- 70.7 GPT-4o
- 70.7
Gemini 2.0 Flash
- 70.7
LongCat-Flash-Omni
- 70.6 MiMo-VL-7B-RL-2508
- 70.4 Claude 3.5 Sonnet (October 2024)
- 70.3
QVQ-72B-Preview
- 70.2
Qwen2.5-VL-72B
- 70.1
InternVL2_5-78B
- 70.1
InternVL3-38B
- 70.0 Kimi k1.5
- 70.0
Magistral Medium 1.0
- 69.4 Grok-3 mini
- 69.4
Llama 4 Scout
- 69.1 GPT-4o
- 69.1
Qwen3-Omni-30B-A3B
- 69.0
Skywork-R1V-38B
- 68.3 Claude 3.5 Sonnet
- 68.0
Gemini 2.0 Flash-Lite
- 68.0
Amazon Nova Premier
- 68.0 GLM-4.1V-9B-Thinking
- 68.0
Nemotron Nano 12B v2 VL
- 67.1
InternVL3-14B
- 66.7 MiMo-VL-7B-RL
- 66.1 Grok-2
- 66.1
Mistral Medium 3
- 66.0
Magistral Small 1.0
- 65.6
Ovis2-34B
- 64.9
Gemma 3 27B
- 64.6 MiMo-VL-7B-SFT
- 64.5 Qwen2-VL-72B-Instruct
- 64.2
Qwen3.5-2B
- 64.0
Mistral Small 3.1
- 64.0 Kimi-VL-A3B-Thinking-2506
- 63.9
InternVL2.5-38B
- 62.7
InternVL3-8B
- 62.5
Mistral Small 3.2
- 62.4
Gemini 1.0 Ultra
- 62.2
Gemini 1.5 Pro
- 62.0
InternVL2-Pro
- 62.0
Amazon Nova Pro
- 61.7 Kimi-VL-A3B-Thinking
- 60.3
Llama 3.2 90B
- 60.0
InternVL2.5-26B
- 59.7
NVLM-D-72B
- 59.6
Ovis2-16B
- 59.6
Gemma 3 12B
- 59.4 Claude 3 Opus
- 59.4 GPT-4o mini
- 59.2
Qwen2.5-Omni-7B
- 59.0
Ovis2-8B
- 58.6
Qwen2.5-VL-7B
- 58.5
Gemini 1.5 Pro
- 58.3
InternVL2-Llama3-76B
- 58.3
Qianfan-VL-70B
- 57.7
InternVL3-9B
- 57.0 Kimi-VL-A3B-Instruct
- 56.8 GPT-4V
- 56.8 LLaVA-OneVision 72B
- 56.3 Reka Core
- 56.1
Gemini 1.5 Flash
- 56.0
InternVL2.5-8B
- 55.8
Eagle2.5-8B
- 55.4 GPT-4.1 nano
- 55.3
BAGEL-7B-MoT
- 55.2
InternVL2-40B
- 55.1
Phi-4-multimodal
- 54.9 Aria
- 54.3
Phi-4-Reasoning-Vision-15B
- 53.3 Reka Flash
- 53.1 Claude 3 Sonnet
- 53.1
Qwen2.5-VL-3B
- 51.6
InternVL-Chat-V1-2
- 51.4 Qwen-VL-Max
- 51.2
InternVL2-8B
- 51.1
DeepSeek-VL2
- 50.8
MAmmoTH-VL-8B
- 50.7
Llama 3.2 11B
- 50.2 Claude 3 Haiku
- 49.9
LLaVA-NeXT-72B
- 49.8
Ovis1.5-Gemma2-9B
- 49.8 MiniCPM-V-2_6
- 49.7
Cambrian-1-34B
- 49.0
Qwen3.5-0.8B
- 48.8
LLaVA-OneVision 7B
- 48.8
Gemma 3 4B
- 48.6
InternVL3-2B
- 48.4 LFM2.5-VL-3B
- 48.2
Llama-3.1-Nemotron-Nano-VL-8B-V1
- 48.1
LLaVA-v1.6-34B
- 48.0
DeepSeek-VL2-Small
- 48.0
Ovis2-4B
- 47.9
Gemini 1.0 Pro
- 47.4 Aquila-VL-2B
- 47.1
Qianfan-VL-8B
- 46.6
Idefics3-8B-Llama3
- 46.4
Qianfan-VL-3B
- 45.9
Yi-VL-34B
- 45.8
MiniCPM-Llama3-V 2.5
- 45.3
Ovis2-2B
- 45.2
Qwen-VL-Plus
- 45.1
Aya Vision 32B
- 43.6
InternVL2.5-2B
- 43.4
InternVL3-1B
- 43.0
InternLM-XComposer2-VL-7B
- 43.0 Phi-3.5-vision-instruct
- 42.8 Reka Edge
- 42.7
Cambrian-1-8B
- 41.4 Bunny-v1.0-4B
- 41.1
OmniLMM-12B
- 41.0
Janus-Pro-7B
- 40.9
InternVL2.5-1B
- 40.7
DeepSeek-VL2-Tiny
- 40.0
Cambrian-1-13B
- 39.9
Aya Vision 8B
- 39.7 LFM2-VL-1.6B
- 39.1
Yi-VL-6B
- 39.1
InternVL-Chat-V1-1
- 38.2 Bunny-v1.0-3B
- 37.2
MiniCPM-V
- 37.1
MiniCPM-V 2.0
- 36.4
LLaVA-v1.5-13B
- 36.3 Emu2-Chat
- 36.2 llava-v1.6-vicuna-13b
- 36.0
Ovis2-1B
- 35.9 Qwen-VL-Chat
- 35.8 llava-v1.6-vicuna-7b
- 35.7
InstructBLIP FLAN-T5-XXL
- 35.4
BLIP-2 FLAN-T5-XXL
- 35.3 llava-v1.6-mistral-7b
- 34.4
BLIP-2 FLAN-T5-XL
- 32.9
InstructBLIP FLAN-T5-XL
- 32.9
North Micro Vision Instruct
- 32.7
mPLUG-Owl2
- 32.1 CogVLM-17B-Chat
- 27.9 Fuyu-8B
- 24.4
Kosmos-2
- 1
Qwen 3.6 Plus 86%
- 2 GPT-5.1 85.4%
- 3
Seed2.0 Pro 85.4%
- 4
Qwen3.5 397B-A17B 85%
- 5 Kimi K2.5 84.3%
- 6 GPT-5 thinking 84.2%
- 7
Gemini 2.5 Deep Think 84%
- 8
Qwen3.5-122B-A10B 83.9%
- 9
Seed1.8 83.4%
- 10 o3 82.9%
- 11
Qwen3.6 27B 82.9%
- 12
Qwen3.5-27B 82.3%
- 13
Gemini 2.5 Pro Jun 2025 82%
- 14
Gemini 2.5 Pro Mar 2025 81.7%
- 15
Qwen 3.6 35B-A3B 81.7%
- 16 o4-mini 81.6%
- 17 GPT-5 mini 81.6%
- 18
Qwen3.5-35B-A3B 81.4%
- 19
gemini-robotics-er-1.5-preview 80.7%
- 20 Claude Opus 4.5 80.7%
- 21
Qwen3-VL-235B-A22B 80.6%
- 22
Gemini 2.5 Flash Sep 2025 80.3%
- 23
MiniCPM-V-4.7-35B-A3B 80.2%
- 24 dots.vlm1 80.1%
- 25
Qwen3.5-Omni-Plus 80.1%
- 26
Gemini 2.5 Flash May 2025 79.7%
- 27
Gemini 2.5 Flash Jun 2025 79.7%
- 28
Gemini 2.5 Pro May 2025 79.6%
- 29
EXAONE 4.5 33B 78.7%
- 30
Qwen3.5-9B 78.4%
- 31 o1 78.2%
- 32
STEP3-VL-10B 78.1%
- 33 Grok 3 78%
- 34
Seed1.5-VL thinking 77.9%
- 35 Claude Sonnet 4.5 77.8%
- 36
Intern-S1 77.7%
- 37
InternVL3.5-241B-A28B 77.7%
- 38
Qwen3.5-4B 77.6%
- 39 Claude Opus 4.1 77.1%
- 40
Gemini 2.5 Flash Apr 2025 76.7%
- 41 Claude Opus 4 76.5%
- 42
Llama 4 Behemoth 76.1%
- 43
Skywork-R1V3-38B 76%
- 44 GLM-4.6V 76%
- 45 GPT-5 nano 75.6%
- 46
Gemini 2.0 Flash Thinking Jan 2025 75.4%
- 47 GLM-4.5V thinking 75.4%
- 48
Keye-VL-2.0-30B-A3B 75.3%
- 49
Command A+ 75.1%
- 50 Claude 3.7 Sonnet 75%
- 51 GPT-4.1 74.8%
- 52
Seed1.6-Thinking 74.8%
- 53 GPT-4.5 74.4%
- 54 Claude Sonnet 4 74.4%
- 55
Step 3 74.2%
- 56
Skywork-R1V2-38B 73.6%
- 57
Llama 4 Maverick 73.4%
- 58 Claude Haiku 4.5 73.2%
- 59
Gemini 2.5 Flash-Lite Jun 2025 72.9%
- 60
Gemini 2.0 Pro 72.7%
- 61 GPT-4.1 mini 72.7%
- 62
InternVL3-78B 72.2%
- 63
ERNIE-4.5-VL-28B-A3B-Thinking 72.2%
- 64
Gemini 2.5 Flash-Lite Sep 2025 72%
- 65
Gemini 2.0 Flash Feb 2025 71.7%
- 66
Nemotron 3 Nano Omni 30B-A3B 70.8%
- 67 GPT-4o Nov 2024 70.7%
- 68
Gemini 2.0 Flash Dec 2024 70.7%
- 69
LongCat-Flash-Omni 70.7%
- 70 MiMo-VL-7B-RL-2508 thinking 70.6%
- 71 Claude 3.5 Sonnet (October 2024) 70.4%
- 72
QVQ-72B-Preview 70.3%
- 73
Qwen2.5-VL-72B 70.2%
- 74
InternVL2_5-78B 70.1%
- 75
InternVL3-38B 70.1%
- 76 Kimi k1.5 70%
- 77
Magistral Medium 1.0 70%
- 78 Grok-3 mini 69.4%
- 79
Llama 4 Scout 69.4%
- 80 GPT-4o May 2024 69.1%
- 81
Qwen3-Omni-30B-A3B 69.1%
- 82
Skywork-R1V-38B 69%
- 83 Claude 3.5 Sonnet 68.3%
- 84
Gemini 2.0 Flash-Lite 68%
- 85
Amazon Nova Premier 68%
- 86 GLM-4.1V-9B-Thinking thinking 68%
- 87
Nemotron Nano 12B v2 VL 68%
- 88
InternVL3-14B 67.1%
- 89 MiMo-VL-7B-RL 66.7%
- 90 Grok-2 Aug 2024 66.1%
- 91
Mistral Medium 3 66.1%
- 92
Magistral Small 1.0 66%
- 93
Ovis2-34B 65.6%
- 94
Gemma 3 27B 64.9%
- 95 MiMo-VL-7B-SFT 64.6%
- 96 Qwen2-VL-72B-Instruct 64.5%
- 97
Qwen3.5-2B 64.2%
- 98
Mistral Small 3.1 64%
- 99 Kimi-VL-A3B-Thinking-2506 64%
- 100
InternVL2.5-38B 63.9%
- 101
InternVL3-8B 62.7%
- 102
Mistral Small 3.2 62.5%
- 103
Gemini 1.0 Ultra 62.4%
- 104
Gemini 1.5 Pro May 2024 62.2%
- 105
InternVL2-Pro 62%
- 106
Amazon Nova Pro 62%
- 107 Kimi-VL-A3B-Thinking 61.7%
- 108
Llama 3.2 90B 60.3%
- 109
InternVL2.5-26B 60%
- 110
NVLM-D-72B 59.7%
- 111
Ovis2-16B 59.6%
- 112
Gemma 3 12B 59.6%
- 113 Claude 3 Opus 59.4%
- 114 GPT-4o mini 59.4%
- 115
Qwen2.5-Omni-7B 59.2%
- 116
Ovis2-8B 59%
- 117
Qwen2.5-VL-7B 58.6%
- 118
Gemini 1.5 Pro Feb 2024 58.5%
- 119
InternVL2-Llama3-76B 58.3%
- 120
Qianfan-VL-70B 58.3%
- 121
InternVL3-9B 57.7%
- 122 Kimi-VL-A3B-Instruct 57%
- 123 GPT-4V 56.8%
- 124 LLaVA-OneVision 72B 56.8%
- 125 Reka Core 56.3%
- 126
Gemini 1.5 Flash May 2024 56.1%
- 127
InternVL2.5-8B 56%
- 128
Eagle2.5-8B 55.8%
- 129 GPT-4.1 nano 55.4%
- 130
BAGEL-7B-MoT 55.3%
- 131
InternVL2-40B 55.2%
- 132
Phi-4-multimodal 55.1%
- 133 Aria 54.9%
- 134
Phi-4-Reasoning-Vision-15B 54.3%
- 135 Reka Flash 53.3%
- 136 Claude 3 Sonnet 53.1%
- 137
Qwen2.5-VL-3B 53.1%
- 138
InternVL-Chat-V1-2 51.6%
- 139 Qwen-VL-Max 51.4%
- 140
InternVL2-8B 51.2%
- 141
DeepSeek-VL2 51.1%
- 142
MAmmoTH-VL-8B 50.8%
- 143
Llama 3.2 11B 50.7%
- 144 Claude 3 Haiku 50.2%
- 145
LLaVA-NeXT-72B 49.9%
- 146
Ovis1.5-Gemma2-9B 49.8%
- 147 MiniCPM-V-2_6 49.8%
- 148
Cambrian-1-34B 49.7%
- 149
Qwen3.5-0.8B 49%
- 150
LLaVA-OneVision 7B 48.8%
- 151
Gemma 3 4B 48.8%
- 152
InternVL3-2B 48.6%
- 153 LFM2.5-VL-3B 48.4%
- 154
Llama-3.1-Nemotron-Nano-VL-8B-V1 48.2%
- 155
LLaVA-v1.6-34B 48.1%
- 156
DeepSeek-VL2-Small 48%
- 157
Ovis2-4B 48%
- 158
Gemini 1.0 Pro 47.9%
- 159 Aquila-VL-2B 47.4%
- 160
Qianfan-VL-8B 47.1%
- 161
Idefics3-8B-Llama3 46.6%
- 162
Qianfan-VL-3B 46.4%
- 163
Yi-VL-34B 45.9%
- 164
MiniCPM-Llama3-V 2.5 45.8%
- 165
Ovis2-2B 45.3%
- 166
Qwen-VL-Plus 45.2%
- 167
Aya Vision 32B 45.1%
- 168
InternVL2.5-2B 43.6%
- 169
InternVL3-1B 43.4%
- 170
InternLM-XComposer2-VL-7B 43%
- 171 Phi-3.5-vision-instruct 43%
- 172 Reka Edge 42.8%
- 173
Cambrian-1-8B 42.7%
- 174 Bunny-v1.0-4B 41.4%
- 175
OmniLMM-12B 41.1%
- 176
Janus-Pro-7B 41%
- 177
InternVL2.5-1B 40.9%
- 178
DeepSeek-VL2-Tiny 40.7%
- 179
Cambrian-1-13B 40%
- 180
Aya Vision 8B 39.9%
- 181 LFM2-VL-1.6B 39.7%
- 182
Yi-VL-6B 39.1%
- 183
InternVL-Chat-V1-1 39.1%
- 184 Bunny-v1.0-3B 38.2%
- 185
MiniCPM-V 37.2%
- 186
MiniCPM-V 2.0 37.1%
- 187
LLaVA-v1.5-13B 36.4%
- 188 Emu2-Chat 36.3%
- 189 llava-v1.6-vicuna-13b 36.2%
- 190
Ovis2-1B 36%
- 191 Qwen-VL-Chat 35.9%
- 192 llava-v1.6-vicuna-7b 35.8%
- 193
InstructBLIP FLAN-T5-XXL 35.7%
- 194
BLIP-2 FLAN-T5-XXL 35.4%
- 195 llava-v1.6-mistral-7b 35.3%
- 196
BLIP-2 FLAN-T5-XL 34.4%
- 197
InstructBLIP FLAN-T5-XL 32.9%
- 198
North Micro Vision Instruct 32.9%
- 199
mPLUG-Owl2 32.7%
- 200 CogVLM-17B-Chat 32.1%
- 201 Fuyu-8B 27.9%
- 202
Kosmos-2 24.4%
Sources: Labs’ model cards and official MMMU leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 86% | Apr 2026 | Lab model card | |
| 2 | GPT-5.1 | 85.4% | Nov 2025 | Lab model card | |
| 3 | | 85.4% | Feb 2026 | Lab model card | |
| 4 | | 85% | Feb 2026 | Lab model card | |
| 5 | Kimi K2.5 | 84.3% | Jan 2026 | Lab model card | |
| 6 | GPT-5 | 84.2% | Aug 2025 | Lab model card | |
| 7 | | 83.9% | Feb 2026 | Lab model card | |
| 8 | | 83.4% | Dec 2025 | Lab model card | |
| 9 | o3 | 82.9% | Apr 2025 | Lab model card | |
| 10 | | 82.9% | Apr 2026 | Lab model card | |
| 11 | | 82.3% | Feb 2026 | Lab model card | |
| 12 | | 82% | Jun 2025 | Lab model card | |
| 13 | | 81.7% | Mar 2025 | Lab model card | |
| 14 | | 81.7% | Apr 2026 | Lab model card | |
| 15 | o4-mini | 81.6% | Apr 2025 | Lab model card | |
| 16 | GPT-5 mini | 81.6% | Aug 2025 | Lab model card | |
| 17 | | 81.4% | Feb 2026 | Lab model card | |
| 18 | | 80.7% | Sep 2025 | Lab model card | |
| 19 | Claude Opus 4.5 | 80.7% | Nov 2025 | Lab model card | |
| 20 | | 80.6% | Sep 2025 | Lab model card | |
| 21 | | 80.3% | Sep 2025 | Lab model card | |
| 22 | | 80.2% | Oct 2026 | Lab model card | |
| 23 | | 80.1% | Mar 2026 | Lab model card | |
| 24 | | 79.7% | Jun 2025 | Lab model card | |
| 25 | | 78.7% | Apr 2026 | Lab model card | |
| 26 | | 78.4% | Mar 2026 | Lab model card | |
| 27 | | 78.1% | Jan 2026 | Lab model card | |
| 28 | Grok 3 | 78% | Feb 2025 | Lab model card | |
| 29 | | 77.9% | May 2025 | Lab model card | |
| 30 | Claude Sonnet 4.5 | 77.8% | Sep 2025 | Lab model card | |
| 31 | | 77.7% | Jul 2025 | Lab model card | |
| 32 | | 77.7% | Aug 2025 | Lab model card | |
| 33 | | 77.6% | Mar 2026 | Lab model card | |
| 34 | o1 | 77.3% | Dec 2024 | Lab model card | |
| 35 | Claude Opus 4.1 | 77.1% | Aug 2025 | Lab model card | |
| 36 | | 76.7% | Apr 2025 | Lab model card | |
| 37 | Claude Opus 4 | 76.5% | May 2025 | Lab model card | |
| 38 | | 76.1% | Apr 2025 | Lab model card | |
| 39 | GLM-4.6V | 76% | Dec 2025 | Lab model card | |
| 40 | GPT-5 nano | 75.6% | Aug 2025 | Lab model card | |
| 41 | | 75.4% | Jan 2025 | Lab model card | |
| 42 | GLM-4.5V | 75.4% | Aug 2025 | Lab model card | |
| 43 | | 75.3% | May 2026 | Lab model card | |
| 44 | | 75.1% | May 2026 | Lab model card | |
| 45 | Claude 3.7 Sonnet | 75% | Feb 2025 | Lab model card | |
| 46 | GPT-4.1 | 74.8% | Apr 2025 | Lab model card | |
| 47 | | 74.8% | Jun 2025 | Lab model card | |
| 48 | GPT-4.5 | 74.4% | Feb 2025 | Lab model card | |
| 49 | Claude Sonnet 4 | 74.4% | May 2025 | Lab model card | |
| 50 | | 74.2% | Jul 2025 | Lab model card | |
| 51 | | 73.4% | Apr 2025 | Lab model card | |
| 52 | Claude Haiku 4.5 | 73.2% | Oct 2025 | Lab model card | |
| 53 | | 72.9% | Jun 2025 | Lab model card | |
| 54 | | 72.7% | Feb 2025 | Lab model card | |
| 55 | GPT-4.1 mini | 72.7% | Apr 2025 | Lab model card | |
| 56 | | 72.2% | Nov 2025 | Lab model card | |
| 57 | | 72% | Sep 2025 | Lab model card | |
| 58 | | 71.7% | Feb 2025 | Lab model card | |
| 59 | | 70.8% | Apr 2026 | Lab model card | |
| 60 | | 70.7% | Dec 2024 | Lab model card | |
| 61 | | 70.7% | Oct 2025 | Lab model card | |
| 62 | Claude 3.5 Sonnet (October 2024) | 70.4% | Oct 2024 | Lab model card | |
| 63 | | 70.2% | Jan 2025 | Lab model card | |
| 64 | Kimi k1.5 | 70% | Jan 2025 | Lab model card | |
| 65 | | 70% | Jun 2025 | Lab model card | |
| 66 | Grok-3 mini | 69.4% | Feb 2025 | Lab model card | |
| 67 | | 69.4% | Apr 2025 | Lab model card | |
| 68 | GPT-4o May 2024 | 69.1% | May 2024 | Lab model card | |
| 69 | | 69.1% | Sep 2025 | Lab model card | |
| 70 | Claude 3.5 Sonnet | 68.3% | Jun 2024 | Lab model card | |
| 71 | | 68% | Feb 2025 | Lab model card | |
| 72 | | 68% | Apr 2025 | Lab model card | |
| 73 | GLM-4.1V-9B-Thinking | 68% | Jul 2025 | Lab model card | |
| 74 | | 68% | Oct 2025 | Lab model card | |
| 75 | Grok-2 Aug 2024 | 66.1% | Aug 2024 | Lab model card | |
| 76 | | 66.1% | May 2025 | Lab model card | |
| 77 | | 66% | Jun 2025 | Lab model card | |
| 78 | | 64.9% | Mar 2025 | Lab model card | |
| 79 | | 64.2% | Mar 2026 | Lab model card | |
| 80 | | 64% | Mar 2025 | Lab model card | |
| 81 | Kimi-VL-A3B-Thinking-2506 | 64% | Jun 2025 | Lab model card | |
| 82 | | 62.5% | Jun 2025 | Lab model card | |
| 83 | | 62.4% | Dec 2023 | Lab model card | |
| 84 | | 62.2% | May 2024 | Lab model card | |
| 85 | Kimi-VL-A3B-Thinking | 61.7% | Apr 2025 | Lab model card | |
| 86 | | 59.6% | Mar 2025 | Lab model card | |
| 87 | Claude 3 Opus | 59.4% | Mar 2024 | Lab model card | |
| 88 | GPT-4o mini | 59.4% | Jul 2024 | Lab model card | |
| 89 | | 59.2% | Mar 2025 | Lab model card | |
| 90 | | 58.5% | Feb 2024 | Lab model card | |
| 91 | GPT-4.1 nano | 55.4% | Apr 2025 | Lab model card | |
| 92 | | 55.3% | May 2025 | Lab model card | |
| 93 | | 55.1% | Feb 2025 | Lab model card | |
| 94 | | 54.3% | Mar 2026 | Lab model card | |
| 95 | Claude 3 Sonnet | 53.1% | Mar 2024 | Lab model card | |
| 96 | Claude 3 Haiku | 50.2% | Mar 2024 | Lab model card | |
| 97 | | 49% | Mar 2026 | Lab model card | |
| 98 | | 48.8% | Mar 2025 | Lab model card | |
| 99 | LFM2.5-VL-3B | 48.4% | Aug 2026 | Lab model card | |
| 100 | | 48.2% | Jun 2025 | Lab model card | |
| 101 | Phi-3.5-vision-instruct | 43% | Aug 2024 | Lab model card | |
| 102 | | 41% | Jan 2025 | Lab model card | |
| 103 | LFM2-VL-1.6B | 39.7% | Nov 2025 | Lab model card | |
| 104 | | 32.9% | Aug 2026 | Lab model card |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.1 | 85.4% | Nov 2025 | Official leaderboard | |
| 2 | GPT-5 | 84.2% | thinking | Aug 2025 | Official leaderboard |
| 3 | 84% | Aug 2025 | Official leaderboard | ||
| 4 | o3 | 82.9% | Apr 2025 | Official leaderboard | |
| 5 | o4-mini | 81.6% | Apr 2025 | Official leaderboard | |
| 6 | Claude Opus 4.5 | 80.7% | Nov 2025 | Official leaderboard | |
| 7 | dots.vlm1 | 80.1% | Aug 2025 | Official leaderboard | |
| 8 | 79.7% | May 2025 | Official leaderboard | ||
| 9 | 79.6% | May 2025 | Official leaderboard | ||
| 10 | 78.7% | Sep 2025 | Official leaderboard | ||
| 11 | o1 | 78.2% | Dec 2024 | Official leaderboard | |
| 12 | Grok 3 | 78% | Apr 2025 | Official leaderboard | |
| 13 | 77.9% | thinking | May 2025 | Official leaderboard | |
| 14 | Claude Sonnet 4.5 | 77.8% | Sep 2025 | Official leaderboard | |
| 15 | 77.7% | Jul 2025 | Official leaderboard | ||
| 16 | Claude Opus 4.1 | 77.1% | Aug 2025 | Official leaderboard | |
| 17 | Claude Opus 4 | 76.5% | May 2025 | Official leaderboard | |
| 18 | 76.1% | Apr 2025 | Official leaderboard | ||
| 19 | 76% | Jul 2025 | Official leaderboard | ||
| 20 | GLM-4.5V | 75.4% | thinking | Aug 2025 | Official leaderboard |
| 21 | Claude 3.7 Sonnet | 75% | Feb 2025 | Official leaderboard | |
| 22 | 74.8% | Jun 2025 | Official leaderboard | ||
| 23 | GPT-4.5 | 74.4% | Feb 2025 | Official leaderboard | |
| 24 | Claude Sonnet 4 | 74.4% | May 2025 | Official leaderboard | |
| 25 | 73.6% | Apr 2025 | Official leaderboard | ||
| 26 | 73.4% | Apr 2025 | Official leaderboard | ||
| 27 | 72.7% | Feb 2025 | Official leaderboard | ||
| 28 | 72.2% | Apr 2025 | Official leaderboard | ||
| 29 | 71.7% | Feb 2025 | Official leaderboard | ||
| 30 | GPT-4o Nov 2024 | 70.7% | Nov 2024 | Official leaderboard | |
| 31 | MiMo-VL-7B-RL-2508 | 70.6% | thinking | Aug 2025 | Official leaderboard |
| 32 | 70.3% | Dec 2024 | Official leaderboard | ||
| 33 | 70.1% | Dec 2024 | Official leaderboard | ||
| 34 | 70.1% | Apr 2025 | Official leaderboard | ||
| 35 | 69.4% | Apr 2025 | Official leaderboard | ||
| 36 | GPT-4o May 2024 | 69.1% | May 2024 | Official leaderboard | |
| 37 | 69% | Mar 2025 | Official leaderboard | ||
| 38 | Claude 3.5 Sonnet | 68.3% | Jun 2024 | Official leaderboard | |
| 39 | 68% | Feb 2025 | Official leaderboard | ||
| 40 | 68% | Apr 2025 | Official leaderboard | ||
| 41 | GLM-4.1V-9B-Thinking | 68% | thinking | Jul 2025 | Official leaderboard |
| 42 | 67.1% | Apr 2025 | Official leaderboard | ||
| 43 | MiMo-VL-7B-RL | 66.7% | May 2025 | Official leaderboard | |
| 44 | 65.6% | Feb 2025 | Official leaderboard | ||
| 45 | 64.9% | Mar 2025 | Official leaderboard | ||
| 46 | MiMo-VL-7B-SFT | 64.6% | May 2025 | Official leaderboard | |
| 47 | Qwen2-VL-72B-Instruct | 64.5% | Aug 2024 | Official leaderboard | |
| 48 | 64.5% | Jan 2025 | Official leaderboard | ||
| 49 | 63.9% | Dec 2024 | Official leaderboard | ||
| 50 | 62.7% | Apr 2025 | Official leaderboard | ||
| 51 | 62.2% | May 2024 | Official leaderboard | ||
| 52 | 62% | Jul 2024 | Official leaderboard | ||
| 53 | 62% | Dec 2024 | Official leaderboard | ||
| 54 | 60.3% | Sep 2024 | Official leaderboard | ||
| 55 | 60% | Dec 2024 | Official leaderboard | ||
| 56 | 59.7% | Sep 2024 | Official leaderboard | ||
| 57 | 59.6% | Feb 2025 | Official leaderboard | ||
| 58 | 59.6% | Mar 2025 | Official leaderboard | ||
| 59 | 59.4% | Dec 2023 | Official leaderboard | ||
| 60 | Claude 3 Opus | 59.4% | Feb 2024 | Official leaderboard | |
| 61 | GPT-4o mini | 59.4% | Jul 2024 | Official leaderboard | |
| 62 | 59% | Feb 2025 | Official leaderboard | ||
| 63 | 58.6% | Jan 2025 | Official leaderboard | ||
| 64 | 58.3% | Jul 2024 | Official leaderboard | ||
| 65 | 58.3% | Sep 2025 | Official leaderboard | ||
| 66 | 57.7% | Apr 2025 | Official leaderboard | ||
| 67 | Kimi-VL-A3B-Instruct | 57% | Apr 2025 | Official leaderboard | |
| 68 | GPT-4V | 56.8% | Nov 2023 | Official leaderboard | |
| 69 | LLaVA-OneVision 72B | 56.8% | Aug 2024 | Official leaderboard | |
| 70 | Reka Core | 56.3% | Apr 2024 | Official leaderboard | |
| 71 | 56.1% | May 2024 | Official leaderboard | ||
| 72 | 56% | Dec 2024 | Official leaderboard | ||
| 73 | 55.8% | Apr 2025 | Official leaderboard | ||
| 74 | 55.2% | Jul 2024 | Official leaderboard | ||
| 75 | Aria | 54.9% | Sep 2024 | Official leaderboard | |
| 76 | Reka Flash | 53.3% | Feb 2024 | Official leaderboard | |
| 77 | Claude 3 Sonnet | 53.1% | Feb 2024 | Official leaderboard | |
| 78 | 53.1% | Jan 2025 | Official leaderboard | ||
| 79 | 51.6% | Feb 2024 | Official leaderboard | ||
| 80 | Qwen-VL-Max | 51.4% | Jan 2024 | Official leaderboard | |
| 81 | 51.2% | Jul 2024 | Official leaderboard | ||
| 82 | 51.1% | Dec 2024 | Official leaderboard | ||
| 83 | 50.8% | Nov 2024 | Official leaderboard | ||
| 84 | 50.7% | Sep 2024 | Official leaderboard | ||
| 85 | Claude 3 Haiku | 50.2% | Mar 2024 | Official leaderboard | |
| 86 | 49.9% | May 2024 | Official leaderboard | ||
| 87 | 49.8% | Jul 2024 | Official leaderboard | ||
| 88 | MiniCPM-V-2_6 | 49.8% | Aug 2024 | Official leaderboard | |
| 89 | 49.7% | Jun 2024 | Official leaderboard | ||
| 90 | 48.8% | Aug 2024 | Official leaderboard | ||
| 91 | 48.8% | Mar 2025 | Official leaderboard | ||
| 92 | 48.6% | Apr 2025 | Official leaderboard | ||
| 93 | 48.1% | Jan 2024 | Official leaderboard | ||
| 94 | 48% | Dec 2024 | Official leaderboard | ||
| 95 | 48% | Feb 2025 | Official leaderboard | ||
| 96 | 47.9% | Dec 2023 | Official leaderboard | ||
| 97 | Aquila-VL-2B | 47.4% | Oct 2024 | Official leaderboard | |
| 98 | 47.1% | Sep 2025 | Official leaderboard | ||
| 99 | 46.6% | Aug 2024 | Official leaderboard | ||
| 100 | 46.4% | Sep 2025 | Official leaderboard | ||
| 101 | 45.9% | Dec 2023 | Official leaderboard | ||
| 102 | 45.8% | May 2024 | Official leaderboard | ||
| 103 | 45.3% | Feb 2025 | Official leaderboard | ||
| 104 | 45.2% | Jan 2024 | Official leaderboard | ||
| 105 | 45.1% | Mar 2025 | Official leaderboard | ||
| 106 | 43.6% | Dec 2024 | Official leaderboard | ||
| 107 | 43.4% | Apr 2025 | Official leaderboard | ||
| 108 | 43% | Jan 2024 | Official leaderboard | ||
| 109 | Phi-3.5-vision-instruct | 43% | Aug 2024 | Official leaderboard | |
| 110 | Reka Edge | 42.8% | Mar 2026 | Official leaderboard | |
| 111 | 42.7% | Jun 2024 | Official leaderboard | ||
| 112 | Bunny-v1.0-4B | 41.4% | Apr 2024 | Official leaderboard | |
| 113 | 41.1% | Jan 2024 | Official leaderboard | ||
| 114 | 40.9% | Dec 2024 | Official leaderboard | ||
| 115 | 40.7% | Dec 2024 | Official leaderboard | ||
| 116 | 40% | Jun 2024 | Official leaderboard | ||
| 117 | 39.9% | Mar 2025 | Official leaderboard | ||
| 118 | 39.1% | Dec 2023 | Official leaderboard | ||
| 119 | 39.1% | Jan 2024 | Official leaderboard | ||
| 120 | Bunny-v1.0-3B | 38.2% | Feb 2024 | Official leaderboard | |
| 121 | 37.2% | Jan 2024 | Official leaderboard | ||
| 122 | 37.1% | Apr 2024 | Official leaderboard | ||
| 123 | 36.4% | Oct 2023 | Official leaderboard | ||
| 124 | Emu2-Chat | 36.3% | Dec 2023 | Official leaderboard | |
| 125 | llava-v1.6-vicuna-13b | 36.2% | Jan 2024 | Official leaderboard | |
| 126 | 36% | Feb 2025 | Official leaderboard | ||
| 127 | Qwen-VL-Chat | 35.9% | Aug 2023 | Official leaderboard | |
| 128 | llava-v1.6-vicuna-7b | 35.8% | Jan 2024 | Official leaderboard | |
| 129 | 35.7% | Jun 2023 | Official leaderboard | ||
| 130 | 35.4% | Feb 2023 | Official leaderboard | ||
| 131 | llava-v1.6-mistral-7b | 35.3% | Jan 2024 | Official leaderboard | |
| 132 | 34.4% | Feb 2023 | Official leaderboard | ||
| 133 | 32.9% | May 2023 | Official leaderboard | ||
| 134 | 32.7% | Nov 2023 | Official leaderboard | ||
| 135 | CogVLM-17B-Chat | 32.1% | Nov 2023 | Official leaderboard | |
| 136 | Fuyu-8B | 27.9% | Oct 2023 | Official leaderboard | |
| 137 | 24.4% | Jun 2023 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | 86% | Apr 2026 | Lab model card | ||
| 2 | GPT-5.1 | 85.4% | Nov 2025 | Official leaderboard | |
| 3 | 85.4% | Feb 2026 | Lab model card | ||
| 4 | 85% | Feb 2026 | Lab model card | ||
| 5 | Kimi K2.5 | 84.3% | Jan 2026 | Lab model card | |
| 6 | GPT-5 | 84.2% | thinking | Aug 2025 | Official leaderboard |
| 7 | 84% | Aug 2025 | Official leaderboard | ||
| 8 | 83.9% | Feb 2026 | Lab model card | ||
| 9 | 83.4% | Dec 2025 | Lab model card | ||
| 10 | o3 | 82.9% | Apr 2025 | Official leaderboard | |
| 11 | 82.9% | Apr 2026 | Lab model card | ||
| 12 | 82.3% | Feb 2026 | Lab model card | ||
| 13 | 82% | Jun 2025 | Lab model card | ||
| 14 | 81.7% | Mar 2025 | Lab model card | ||
| 15 | 81.7% | Apr 2026 | Lab model card | ||
| 16 | o4-mini | 81.6% | Apr 2025 | Official leaderboard | |
| 17 | GPT-5 mini | 81.6% | Aug 2025 | Lab model card | |
| 18 | 81.4% | Feb 2026 | Lab model card | ||
| 19 | 80.7% | Sep 2025 | Lab model card | ||
| 20 | Claude Opus 4.5 | 80.7% | Nov 2025 | Official leaderboard | |
| 21 | 80.6% | Sep 2025 | Lab model card | ||
| 22 | 80.3% | Sep 2025 | Lab model card | ||
| 23 | 80.2% | Oct 2026 | Lab model card | ||
| 24 | dots.vlm1 | 80.1% | Aug 2025 | Official leaderboard | |
| 25 | 80.1% | Mar 2026 | Lab model card | ||
| 26 | 79.7% | May 2025 | Official leaderboard | ||
| 27 | 79.7% | Jun 2025 | Lab model card | ||
| 28 | 79.6% | May 2025 | Official leaderboard | ||
| 29 | 78.7% | Apr 2026 | Lab model card | ||
| 30 | 78.4% | Mar 2026 | Lab model card | ||
| 31 | o1 | 78.2% | Dec 2024 | Official leaderboard | |
| 32 | 78.1% | Jan 2026 | Lab model card | ||
| 33 | Grok 3 | 78% | Apr 2025 | Official leaderboard | |
| 34 | 77.9% | thinking | May 2025 | Official leaderboard | |
| 35 | Claude Sonnet 4.5 | 77.8% | Sep 2025 | Official leaderboard | |
| 36 | 77.7% | Jul 2025 | Official leaderboard | ||
| 37 | 77.7% | Aug 2025 | Lab model card | ||
| 38 | 77.6% | Mar 2026 | Lab model card | ||
| 39 | Claude Opus 4.1 | 77.1% | Aug 2025 | Official leaderboard | |
| 40 | 76.7% | Apr 2025 | Lab model card | ||
| 41 | Claude Opus 4 | 76.5% | May 2025 | Official leaderboard | |
| 42 | 76.1% | Apr 2025 | Official leaderboard | ||
| 43 | 76% | Jul 2025 | Official leaderboard | ||
| 44 | GLM-4.6V | 76% | Dec 2025 | Lab model card | |
| 45 | GPT-5 nano | 75.6% | Aug 2025 | Lab model card | |
| 46 | 75.4% | Jan 2025 | Lab model card | ||
| 47 | GLM-4.5V | 75.4% | thinking | Aug 2025 | Official leaderboard |
| 48 | 75.3% | May 2026 | Lab model card | ||
| 49 | 75.1% | May 2026 | Lab model card | ||
| 50 | Claude 3.7 Sonnet | 75% | Feb 2025 | Official leaderboard | |
| 51 | GPT-4.1 | 74.8% | Apr 2025 | Lab model card | |
| 52 | 74.8% | Jun 2025 | Official leaderboard | ||
| 53 | GPT-4.5 | 74.4% | Feb 2025 | Official leaderboard | |
| 54 | Claude Sonnet 4 | 74.4% | May 2025 | Official leaderboard | |
| 55 | 74.2% | Jul 2025 | Lab model card | ||
| 56 | 73.6% | Apr 2025 | Official leaderboard | ||
| 57 | 73.4% | Apr 2025 | Official leaderboard | ||
| 58 | Claude Haiku 4.5 | 73.2% | Oct 2025 | Lab model card | |
| 59 | 72.9% | Jun 2025 | Lab model card | ||
| 60 | 72.7% | Feb 2025 | Official leaderboard | ||
| 61 | GPT-4.1 mini | 72.7% | Apr 2025 | Lab model card | |
| 62 | 72.2% | Apr 2025 | Official leaderboard | ||
| 63 | 72.2% | Nov 2025 | Lab model card | ||
| 64 | 72% | Sep 2025 | Lab model card | ||
| 65 | 71.7% | Feb 2025 | Official leaderboard | ||
| 66 | 70.8% | Apr 2026 | Lab model card | ||
| 67 | GPT-4o Nov 2024 | 70.7% | Nov 2024 | Official leaderboard | |
| 68 | 70.7% | Dec 2024 | Lab model card | ||
| 69 | 70.7% | Oct 2025 | Lab model card | ||
| 70 | MiMo-VL-7B-RL-2508 | 70.6% | thinking | Aug 2025 | Official leaderboard |
| 71 | Claude 3.5 Sonnet (October 2024) | 70.4% | Oct 2024 | Lab model card | |
| 72 | 70.3% | Dec 2024 | Official leaderboard | ||
| 73 | 70.2% | Jan 2025 | Lab model card | ||
| 74 | 70.1% | Dec 2024 | Official leaderboard | ||
| 75 | 70.1% | Apr 2025 | Official leaderboard | ||
| 76 | Kimi k1.5 | 70% | Jan 2025 | Lab model card | |
| 77 | 70% | Jun 2025 | Lab model card | ||
| 78 | Grok-3 mini | 69.4% | Feb 2025 | Lab model card | |
| 79 | 69.4% | Apr 2025 | Official leaderboard | ||
| 80 | GPT-4o May 2024 | 69.1% | May 2024 | Official leaderboard | |
| 81 | 69.1% | Sep 2025 | Lab model card | ||
| 82 | 69% | Mar 2025 | Official leaderboard | ||
| 83 | Claude 3.5 Sonnet | 68.3% | Jun 2024 | Official leaderboard | |
| 84 | 68% | Feb 2025 | Official leaderboard | ||
| 85 | 68% | Apr 2025 | Official leaderboard | ||
| 86 | GLM-4.1V-9B-Thinking | 68% | thinking | Jul 2025 | Official leaderboard |
| 87 | 68% | Oct 2025 | Lab model card | ||
| 88 | 67.1% | Apr 2025 | Official leaderboard | ||
| 89 | MiMo-VL-7B-RL | 66.7% | May 2025 | Official leaderboard | |
| 90 | Grok-2 Aug 2024 | 66.1% | Aug 2024 | Lab model card | |
| 91 | 66.1% | May 2025 | Lab model card | ||
| 92 | 66% | Jun 2025 | Lab model card | ||
| 93 | 65.6% | Feb 2025 | Official leaderboard | ||
| 94 | 64.9% | Mar 2025 | Official leaderboard | ||
| 95 | MiMo-VL-7B-SFT | 64.6% | May 2025 | Official leaderboard | |
| 96 | Qwen2-VL-72B-Instruct | 64.5% | Aug 2024 | Official leaderboard | |
| 97 | 64.2% | Mar 2026 | Lab model card | ||
| 98 | 64% | Mar 2025 | Lab model card | ||
| 99 | Kimi-VL-A3B-Thinking-2506 | 64% | Jun 2025 | Lab model card | |
| 100 | 63.9% | Dec 2024 | Official leaderboard | ||
| 101 | 62.7% | Apr 2025 | Official leaderboard | ||
| 102 | 62.5% | Jun 2025 | Lab model card | ||
| 103 | 62.4% | Dec 2023 | Lab model card | ||
| 104 | 62.2% | May 2024 | Official leaderboard | ||
| 105 | 62% | Jul 2024 | Official leaderboard | ||
| 106 | 62% | Dec 2024 | Official leaderboard | ||
| 107 | Kimi-VL-A3B-Thinking | 61.7% | Apr 2025 | Lab model card | |
| 108 | 60.3% | Sep 2024 | Official leaderboard | ||
| 109 | 60% | Dec 2024 | Official leaderboard | ||
| 110 | 59.7% | Sep 2024 | Official leaderboard | ||
| 111 | 59.6% | Feb 2025 | Official leaderboard | ||
| 112 | 59.6% | Mar 2025 | Official leaderboard | ||
| 113 | Claude 3 Opus | 59.4% | Feb 2024 | Official leaderboard | |
| 114 | GPT-4o mini | 59.4% | Jul 2024 | Official leaderboard | |
| 115 | 59.2% | Mar 2025 | Lab model card | ||
| 116 | 59% | Feb 2025 | Official leaderboard | ||
| 117 | 58.6% | Jan 2025 | Official leaderboard | ||
| 118 | 58.5% | Feb 2024 | Lab model card | ||
| 119 | 58.3% | Jul 2024 | Official leaderboard | ||
| 120 | 58.3% | Sep 2025 | Official leaderboard | ||
| 121 | 57.7% | Apr 2025 | Official leaderboard | ||
| 122 | Kimi-VL-A3B-Instruct | 57% | Apr 2025 | Official leaderboard | |
| 123 | GPT-4V | 56.8% | Nov 2023 | Official leaderboard | |
| 124 | LLaVA-OneVision 72B | 56.8% | Aug 2024 | Official leaderboard | |
| 125 | Reka Core | 56.3% | Apr 2024 | Official leaderboard | |
| 126 | 56.1% | May 2024 | Official leaderboard | ||
| 127 | 56% | Dec 2024 | Official leaderboard | ||
| 128 | 55.8% | Apr 2025 | Official leaderboard | ||
| 129 | GPT-4.1 nano | 55.4% | Apr 2025 | Lab model card | |
| 130 | 55.3% | May 2025 | Lab model card | ||
| 131 | 55.2% | Jul 2024 | Official leaderboard | ||
| 132 | 55.1% | Feb 2025 | Lab model card | ||
| 133 | Aria | 54.9% | Sep 2024 | Official leaderboard | |
| 134 | 54.3% | Mar 2026 | Lab model card | ||
| 135 | Reka Flash | 53.3% | Feb 2024 | Official leaderboard | |
| 136 | Claude 3 Sonnet | 53.1% | Feb 2024 | Official leaderboard | |
| 137 | 53.1% | Jan 2025 | Official leaderboard | ||
| 138 | 51.6% | Feb 2024 | Official leaderboard | ||
| 139 | Qwen-VL-Max | 51.4% | Jan 2024 | Official leaderboard | |
| 140 | 51.2% | Jul 2024 | Official leaderboard | ||
| 141 | 51.1% | Dec 2024 | Official leaderboard | ||
| 142 | 50.8% | Nov 2024 | Official leaderboard | ||
| 143 | 50.7% | Sep 2024 | Official leaderboard | ||
| 144 | Claude 3 Haiku | 50.2% | Mar 2024 | Official leaderboard | |
| 145 | 49.9% | May 2024 | Official leaderboard | ||
| 146 | 49.8% | Jul 2024 | Official leaderboard | ||
| 147 | MiniCPM-V-2_6 | 49.8% | Aug 2024 | Official leaderboard | |
| 148 | 49.7% | Jun 2024 | Official leaderboard | ||
| 149 | 49% | Mar 2026 | Lab model card | ||
| 150 | 48.8% | Aug 2024 | Official leaderboard | ||
| 151 | 48.8% | Mar 2025 | Official leaderboard | ||
| 152 | 48.6% | Apr 2025 | Official leaderboard | ||
| 153 | LFM2.5-VL-3B | 48.4% | Aug 2026 | Lab model card | |
| 154 | 48.2% | Jun 2025 | Lab model card | ||
| 155 | 48.1% | Jan 2024 | Official leaderboard | ||
| 156 | 48% | Dec 2024 | Official leaderboard | ||
| 157 | 48% | Feb 2025 | Official leaderboard | ||
| 158 | 47.9% | Dec 2023 | Official leaderboard | ||
| 159 | Aquila-VL-2B | 47.4% | Oct 2024 | Official leaderboard | |
| 160 | 47.1% | Sep 2025 | Official leaderboard | ||
| 161 | 46.6% | Aug 2024 | Official leaderboard | ||
| 162 | 46.4% | Sep 2025 | Official leaderboard | ||
| 163 | 45.9% | Dec 2023 | Official leaderboard | ||
| 164 | 45.8% | May 2024 | Official leaderboard | ||
| 165 | 45.3% | Feb 2025 | Official leaderboard | ||
| 166 | 45.2% | Jan 2024 | Official leaderboard | ||
| 167 | 45.1% | Mar 2025 | Official leaderboard | ||
| 168 | 43.6% | Dec 2024 | Official leaderboard | ||
| 169 | 43.4% | Apr 2025 | Official leaderboard | ||
| 170 | 43% | Jan 2024 | Official leaderboard | ||
| 171 | Phi-3.5-vision-instruct | 43% | Aug 2024 | Official leaderboard | |
| 172 | Reka Edge | 42.8% | Mar 2026 | Official leaderboard | |
| 173 | 42.7% | Jun 2024 | Official leaderboard | ||
| 174 | Bunny-v1.0-4B | 41.4% | Apr 2024 | Official leaderboard | |
| 175 | 41.1% | Jan 2024 | Official leaderboard | ||
| 176 | 41% | Jan 2025 | Lab model card | ||
| 177 | 40.9% | Dec 2024 | Official leaderboard | ||
| 178 | 40.7% | Dec 2024 | Official leaderboard | ||
| 179 | 40% | Jun 2024 | Official leaderboard | ||
| 180 | 39.9% | Mar 2025 | Official leaderboard | ||
| 181 | LFM2-VL-1.6B | 39.7% | Nov 2025 | Lab model card | |
| 182 | 39.1% | Dec 2023 | Official leaderboard | ||
| 183 | 39.1% | Jan 2024 | Official leaderboard | ||
| 184 | Bunny-v1.0-3B | 38.2% | Feb 2024 | Official leaderboard | |
| 185 | 37.2% | Jan 2024 | Official leaderboard | ||
| 186 | 37.1% | Apr 2024 | Official leaderboard | ||
| 187 | 36.4% | Oct 2023 | Official leaderboard | ||
| 188 | Emu2-Chat | 36.3% | Dec 2023 | Official leaderboard | |
| 189 | llava-v1.6-vicuna-13b | 36.2% | Jan 2024 | Official leaderboard | |
| 190 | 36% | Feb 2025 | Official leaderboard | ||
| 191 | Qwen-VL-Chat | 35.9% | Aug 2023 | Official leaderboard | |
| 192 | llava-v1.6-vicuna-7b | 35.8% | Jan 2024 | Official leaderboard | |
| 193 | 35.7% | Jun 2023 | Official leaderboard | ||
| 194 | 35.4% | Feb 2023 | Official leaderboard | ||
| 195 | llava-v1.6-mistral-7b | 35.3% | Jan 2024 | Official leaderboard | |
| 196 | 34.4% | Feb 2023 | Official leaderboard | ||
| 197 | 32.9% | May 2023 | Official leaderboard | ||
| 198 | 32.9% | Aug 2026 | Lab model card | ||
| 199 | 32.7% | Nov 2023 | Official leaderboard | ||
| 200 | CogVLM-17B-Chat | 32.1% | Nov 2023 | Official leaderboard | |
| 201 | Fuyu-8B | 27.9% | Oct 2023 | Official leaderboard | |
| 202 | 24.4% | Jun 2023 | Official leaderboard |
The benchmark's own leaderboard has 186 entries, 137 of them models we track. Switch the scores above to rank them together.
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: Official MMMU leaderboard
11 models have held number one in labs’ own numbers since its release in November 2023 10 models have held number one across all independent runs since its release in November 2023 13 models have held number one with claims included since its release in November 2023
Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.
Sources: Official MMMU leaderboard and labs’ model cards
GPT-5.1 on the official leaderboard passed 85% in November 2025
That was 24 months after its release. By the same age, 43% of the benchmarks from 2023 had passed 85%. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Sources: Official MMMU leaderboard and labs’ model cards
3 labs put it in a launch post in the last six months
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| Oct 7, 2026 | 80.2 | MMMU | |
| Aug 12, 2026 | LFM2.5-VL-3B | 48.4 | MMMU (val) |
| Aug 12, 2026 | 32.9 | MMMU DEV_VAL | |
| May 25, 2026 | 75.3 | MMMU | |
| May 20, 2026 | 75.1 | MMMU (75.1% in text, 75% in chart) | |
| Apr 22, 2026 | 82.9 | MMMU | |
| Apr 15, 2026 | 81.7 | MMMU | |
| Apr 9, 2026 | 78.7 | MMMU | |
| Apr 2, 2026 | 86 | MMMU | |
| Mar 30, 2026 | 80.1 | MMMU | |
| Mar 4, 2026 | 54.3 | MMMU_VAL (default mode) | |
| Mar 2, 2026 | 49 | MMMU 49/47.4 (with CI / without CI) | |
| Mar 2, 2026 | 64.2 | MMMU 64.2/64.2 (with CI / without CI) | |
| Mar 2, 2026 | 77.6 | MMMU | |
| Mar 2, 2026 | 78.4 | MMMU | |
| Feb 24, 2026 | 81.4 | MMMU | |
| Feb 24, 2026 | 82.3 | MMMU | |
| Feb 24, 2026 | 83.9 | MMMU | |
| Feb 16, 2026 | 85 | MMMU | |
| Jan 13, 2026 | 78.1 | MMMU (SeRe) | |
| Dec 18, 2025 | 83.4 | MMMU | |
| Dec 8, 2025 | GLM-4.6V | 76 | MMMU (Val) |
| Nov 24, 2025 | Claude Opus 4.5 | 80.7 | Visual reasoning MMMU (validation) |
| Nov 13, 2025 | GPT-5.1 | 85.4 | MMMU |
| Nov 11, 2025 | 72.2 | MMMU | |
| Oct 31, 2025 | 70.7 | MMMU val | |
| Oct 28, 2025 | 68 | MMMU* (val, ChatGPT judge) | |
| Oct 15, 2025 | Claude Haiku 4.5 | 73.2 | Visual reasoning MMMU (validation) |
| Sep 29, 2025 | Claude Sonnet 4.5 | 77.8 | Visual reasoning MMMU (validation) |
| Sep 25, 2025 | 72 | MMMU single attempt, Thinking | |
| Sep 25, 2025 | 80.3 | MMMU single attempt, model card | |
| Sep 23, 2025 | 80.6 | MMMU_VAL | |
| Sep 22, 2025 | 69.1 | MMMU_val | |
| Aug 25, 2025 | 77.7 | MMMU (val) | |
| Aug 12, 2025 | LFM2-VL-1.6B | 38.4 | MMMU (Val) |
| Aug 11, 2025 | GLM-4.5V | 75.4 | MMMU (val) |
| Aug 7, 2025 | GPT-5 nano | 75.6 | MMMU |
| Aug 7, 2025 | GPT-5 mini | 81.6 | MMMU |
| Aug 7, 2025 | GPT-5 | 84.2 | MMMU College-level visual problem-solving |
| Aug 5, 2025 | Claude Opus 4.1 | 77.1 | Visual reasoning MMMU (validation) |
| Jul 31, 2025 | MMMU (chart only) | ||
| Jul 31, 2025 | 74.2 | MMMU | |
| Jul 24, 2025 | 77.7 | MMMU | |
| Jul 1, 2025 | GLM-4.1V-9B-Thinking | 68 | MMMU |
| Jun 25, 2025 | 74.8 | MMMU | |
| Jun 21, 2025 | Kimi-VL-A3B-Thinking-2506 | 64 | MMMU (val, Pass@1) |
| Jun 20, 2025 | 62.5 | MMMU | |
| Jun 17, 2025 | 79.7 | MMMU single attempt | |
| Jun 17, 2025 | 72.9 | MMMU single attempt, Thinking | |
| Jun 17, 2025 | 82 | MMMU, single attempt (pass@1) | |
| Jun 3, 2025 | 48.2 | MMMU Val with chatGPT as a judge | |
| May 22, 2025 | Claude Sonnet 4 | 74.4 | Visual reasoning MMMU (validation) |
| May 22, 2025 | Claude Opus 4 | 76.5 | Visual reasoning MMMU (validation) |
| May 20, 2025 | 55.3 | MMMU | |
| May 12, 2025 | 77.9 | MMMU | |
| May 7, 2025 | 66.1 | MMMU 0-shot | |
| Apr 30, 2025 | 68 | MMMU | |
| Apr 17, 2025 | 76.7 | MMMU single attempt | |
| Apr 16, 2025 | o4-mini | 81.6 | MMMU |
| Apr 16, 2025 | o3 | 82.9 | MMMU College-level visual problem-solving |
| Apr 14, 2025 | GPT-4.1 nano | 55.4 | MMMU |
| Apr 14, 2025 | GPT-4.1 mini | 72.7 | MMMU |
| Apr 14, 2025 | GPT-4.1 | 74.8 | MMMU |
| Apr 10, 2025 | Kimi-VL-A3B-Thinking | 61.7 | MMMU (val) (Pass@1) |
| Apr 5, 2025 | 69.4 | MMMU, 0-shot, accuracy | |
| Apr 5, 2025 | 73.4 | MMMU, 0-shot, accuracy | |
| Apr 5, 2025 | 76.1 | MMMU | |
| Mar 27, 2025 | 59.2 | MMMUval | |
| Mar 25, 2025 | 81.7 | MMMU, single attempt (pass@1) | |
| Mar 17, 2025 | 64 | MMMU | |
| Mar 16, 2025 | MMMU | ||
| Mar 12, 2025 | 64.9 | MMMU (val) (IT); PT MMMU = 56.1 | |
| Feb 27, 2025 | GPT-4.5 | 74.4 | MMMU (multimodal) |
| Feb 26, 2025 | 55.1 | MMMU | |
| Feb 24, 2025 | Claude 3.7 Sonnet | 75 | Visual reasoning MMMU (validation) (64K extended thinking) |
| Feb 19, 2025 | Grok-3 mini | 69.4 | MMMU [Grok 3 mini, non-reasoning table] |
| Feb 19, 2025 | Grok 3 | 78 | MMMU (Grok 3 Think; pass@1 = 76) |
| Feb 5, 2025 | 68 | MMMU | |
| Feb 5, 2025 | 71.7 | MMMU | |
| Feb 5, 2025 | 72.7 | MMMU | |
| Jan 28, 2025 | 70.2 | MMMUval | |
| Jan 27, 2025 | 41 | MMMU | |
| Jan 21, 2025 | 75.4 | MMMU (Multimodal reasoning) | |
| Jan 20, 2025 | Kimi k1.5 | 70 | MMMU-Val (Pass@1), long-CoT |
| Dec 17, 2024 | o1 | 77.3 | MMMU (pass @1) |
| Dec 11, 2024 | 70.7 | MMMU | |
| Oct 22, 2024 | Claude 3.5 Sonnet (October 2024) | 70.4 | MMMU (validation) |
| Sep 12, 2024 | o1-preview | MMMU (val) pass@1 | |
| Aug 13, 2024 | Grok-2 | 66.1 | MMMU (0-shot CoT) |
| Jul 18, 2024 | GPT-4o mini | 59.4 | MMMU |
| Jun 20, 2024 | Claude 3.5 Sonnet | 68.3 | MMMU (validation) |
| May 14, 2024 | 62.2 | MMMU (val), 0-shot | |
| May 13, 2024 | GPT-4o | 69.1 | MMMU (%) (val) |
| Mar 4, 2024 | Claude 3 Opus | 59.4 | MMMU (val) Overall |
| Feb 15, 2024 | 58.5 | MMMU (val), 0-shot | |
| Dec 6, 2023 | 62.4 | MMMU (val), Maj1@32 (pass@1 = 59.4%) |
51 system cards report it, from 18 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| Jun 9, 2026 | | 75.3 | MMMU (Figure 5, printed value) | Capabilities |
| Apr 27, 2026 | | 70.8 | MMMU (val), reasoning on, 55.2 reasoning off (Table 7) | Capabilities |
| Apr 17, 2026 | | 80.1 | MMMU (Table 6) | Capabilities |
| Apr 9, 2026 | | 78.7 | MMMU (Table 2, Reasoning mode) | Capabilities |
| Mar 21, 2026 | | 83.4 | MMMU (Table 2) | Capabilities |
| Mar 4, 2026 | | 54.3 | MMMU VAL, default mode, 52 force nothink, 55 force thinking (Tables 5 and 6) | Capabilities |
| Feb 14, 2026 | | 85.4 | MMMU (Table 8) | Capabilities |
| Jan 30, 2026 | Kimi K2.5 technical report | 84.3 | MMMU (val) (Table 4) | Capabilities |
| Jan 14, 2026 | | 78.1 | MMMU, SeRe (Tables 1 and 3). PaCoRe 80.11 | Capabilities |
| Jan 1, 2026 | GLM-4.6V technical report | 76 | MMMU (Val), GLM-4.6V thinking (Table 2) | Capabilities |
| Nov 28, 2025 | LFM2-VL-1.6B technical report | 39.7 | MMMU, LFM2-VL-1.6B (Table 8) | Capabilities |
| Nov 26, 2025 | | 80.6 | MMMU (Table 2) | Capabilities |
| Nov 24, 2025 | Claude Opus 4.5 system card | 80.7 | MMMU (validation), 64k thinking, avg of 5 trials (Table 2.3.A; 80.72% in 2.20) | Capabilities |
| Nov 6, 2025 | | 67.8 | MMMU (val), reasoning on (Table 3). Reasoning off 55.3 | Capabilities |
| Oct 31, 2025 | | 70.7 | MMMU val (Table 5) | Capabilities |
| Sep 26, 2025 | | 80.3 | Visual reasoning MMMU, single attempt (pass@1) (Table, p.5) | Capabilities |
| Sep 25, 2025 | | 80.7 | MMMU (Table 20, thinking on). Thinking off 79.3 | Capabilities |
| Sep 22, 2025 | | 69.1 | MMMU val, Instruct (Table 9). Thinking 75.6 (Table 10) | Capabilities |
| Aug 25, 2025 | | 77.7 | MMMU (val) (Table 2, Table 3). With parallel thinking 78.7 | Capabilities |
| Aug 21, 2025 | | 77.7 | MMMU (Table 2) | Capabilities |
| Aug 13, 2025 | GLM-4.5V technical report | 75.4 | MMMU (Val), GLM-4.5V thinking (Table 2) | Capabilities |
| Jul 7, 2025 | | 82 | MMMU, single attempt (Table 3) | Capabilities |
| Jul 7, 2025 | | 79.7 | MMMU, single attempt (Table 3) | Capabilities |
| Jul 1, 2025 | GLM-4.1V-9B-Thinking technical report | 68 | MMMU (Val), GLM-4.1V-9B-Thinking thinking (Table 2) | Capabilities |
| Jun 23, 2025 | Kimi-VL-A3B-Thinking-2506 technical report | 64 | MMMU (val) (Pass@1) (Table 4) | Capabilities |
| Jun 12, 2025 | | 66 | MMMU, Magistral Small bar, printed label (Figure 10) | Capabilities |
| Jun 12, 2025 | | 70 | MMMU, printed bar label (Figure 10) | Capabilities |
| May 20, 2025 | | 55.3 | MMMU (Table 4) | Capabilities |
| May 11, 2025 | | 77.9 | MMMU, thinking (Table 6). Non-thinking 73.6 | Capabilities |
| Apr 30, 2025 | | 68 | MMMU, CoT (Table 1) | Capabilities |
| Apr 29, 2025 | | 76.7 | MMMU, single attempt (pass@1) (Table, p.4) | Capabilities |
| Apr 28, 2025 | | 81.7 | MMMU, single attempt pass@1 (Table, p.4, column Gemini 2.5 Pro Experimental 03-25) | Capabilities |
| Apr 15, 2025 | | 71.7 | MMMU (Table, p.3, column Gemini 2.0 Flash (GA)) | Capabilities |
| Apr 10, 2025 | Kimi-VL-A3B-Thinking technical report | 61.7 | MMMU (val) (Pass@1) (Table 4). Figure 13 shows 49.2 at 1k rising to 61.7 at 16k max thinking tokens | Capabilities |
| Apr 10, 2025 | | 68 | MMMU (Table, p.3) | Capabilities |
| Apr 5, 2025 | | 69.4 | MMMU, 0-shot, accuracy, instruction tuned (Instruction tuned models table) | Capabilities |
| Apr 5, 2025 | | 73.4 | MMMU, 0-shot, accuracy, instruction tuned (Instruction tuned models table) | Capabilities |
| Mar 26, 2025 | | 59.2 | MMMU val (Table 5) | Capabilities |
| Mar 25, 2025 | | 48.8 | MMMU (val), IT (Table 6 and Table 16) | Capabilities |
| Mar 25, 2025 | | 64.9 | MMMU (val), IT (Table 6 and Table 16) | Capabilities |
| Mar 25, 2025 | | 59.6 | MMMU (val), IT (Table 6 and Table 16) | Capabilities |
| Mar 3, 2025 | | 55.1 | MMMU (val) (Table 1) | Capabilities |
| Feb 19, 2025 | | 70.2 | MMMU val (Table 3) | Capabilities |
| Jan 27, 2025 | | 41 | MMMU (Table 3) | Capabilities |
| Jan 20, 2025 | Kimi k1.5 technical report | 70 | MMMU-Val (Pass@1), long-CoT (Table 2). Short-CoT 68.0 (Table 3) | Capabilities |
| Oct 22, 2024 | Claude 3.5 Sonnet (October 2024) model card | 70.4 | MMMU (validation), 0-shot CoT (Table 5) | Capabilities |
| Aug 30, 2024 | | 43 | MMMU (val), 0-shot (Table 5) | Capabilities |
| Jun 20, 2024 | Claude 3.5 Sonnet model card | 68.3 | MMMU (validation), 0-shot CoT (Table 2) | Capabilities |
| Mar 4, 2024 | Claude 3 Sonnet model card | 53.1 | MMMU (val), Overall, 0-shot. Per-discipline values also given (Table 3) | Capabilities |
| Mar 4, 2024 | Claude 3 Opus model card | 59.4 | MMMU (val), Overall, 0-shot. Per-discipline values also given (Table 3) | Capabilities |
| Mar 4, 2024 | Claude 3 Haiku model card | 50.2 | MMMU (val), Overall, 0-shot. Per-discipline values also given (Table 3) | Capabilities |
The lab's number comes out ahead in 3 of 4 pairs that differ
Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 35 of them match, which usually means one side copied the other. Every benchmark.
| Model | Lab says | Board says | Gap |
|---|---|---|---|
| | 70.2% MMMUval | 64.5% Leaderboard | +5.7 pts |
| | 62.4% MMMU (val), Maj1@32 (pass@1 = 59.4%) | 59.4% Leaderboard | +3.0 pts |
| | 80.6% MMMU_VAL | 78.7% Leaderboard | +1.9 pts |
| o1 | 77.3% MMMU (pass @1) | 78.2% Leaderboard | −0.9 pts |
| Claude 3 Opus | 59.4% MMMU (val) Overall | 59.4% Leaderboard | Same |
| GPT-4o May 2024 | 69.1% MMMU (%) (val) | 69.1% Leaderboard | Same |
| | 62.2% MMMU (val), 0-shot | 62.2% Leaderboard | Same |
| Claude 3.5 Sonnet | 68.3% MMMU (validation) | 68.3% Leaderboard | Same |
| GPT-4o mini | 59.4% MMMU | 59.4% Leaderboard | Same |
| | 72.7% MMMU | 72.7% Leaderboard | Same |
| | 71.7% MMMU | 71.7% Leaderboard | Same |
| | 68% MMMU | 68% Leaderboard | Same |
| Grok 3 | 78% MMMU (Grok 3 Think; pass@1 = 76) | 78% Leaderboard | Same |
| Claude 3.7 Sonnet | 75% Visual reasoning MMMU (validation) (64K extended thinking) | 75% Leaderboard | Same |
| GPT-4.5 | 74.4% MMMU (multimodal) | 74.4% Leaderboard | Same |
| | 64.9% MMMU (val) (IT); PT MMMU = 56.1 | 64.9% Leaderboard | Same |
| | 76.1% MMMU | 76.1% Leaderboard | Same |
| | 73.4% MMMU, 0-shot, accuracy | 73.4% Leaderboard | Same |
| | 69.4% MMMU, 0-shot, accuracy | 69.4% Leaderboard | Same |
| o3 | 82.9% MMMU College-level visual problem-solving | 82.9% Leaderboard | Same |
| o4-mini | 81.6% MMMU | 81.6% Leaderboard | Same |
| | 68% MMMU | 68% Leaderboard | Same |
| | 77.9% MMMU | 77.9% Leaderboard, thinking | Same |
| Claude Opus 4 | 76.5% Visual reasoning MMMU (validation) | 76.5% Leaderboard | Same |
| Claude Sonnet 4 | 74.4% Visual reasoning MMMU (validation) | 74.4% Leaderboard | Same |
| | 74.8% MMMU | 74.8% Leaderboard | Same |
| GLM-4.1V-9B-Thinking | 68% MMMU | 68% Leaderboard, thinking | Same |
| | 77.7% MMMU | 77.7% Leaderboard | Same |
| Claude Opus 4.1 | 77.1% Visual reasoning MMMU (validation) | 77.1% Leaderboard | Same |
| GPT-5 | 84.2% MMMU College-level visual problem-solving | 84.2% Leaderboard, thinking | Same |
| GLM-4.5V | 75.4% MMMU (val) | 75.4% Leaderboard, thinking | Same |
| Claude Sonnet 4.5 | 77.8% Visual reasoning MMMU (validation) | 77.8% Leaderboard | Same |
| GPT-5.1 | 85.4% MMMU | 85.4% Leaderboard | Same |
| Claude Opus 4.5 | 80.7% Visual reasoning MMMU (validation) | 80.7% Leaderboard | Same |
| Claude 3 Haiku | 50.2% MMMU (val), Overall, 0-shot. Per-discipline values also given (Table 3) | 50.2% Leaderboard | Same |
| Claude 3 Sonnet | 53.1% MMMU (val), Overall, 0-shot. Per-discipline values also given (Table 3) | 53.1% Leaderboard | Same |
| Phi-3.5-vision-instruct | 43% MMMU (val), 0-shot (Table 5) | 43% Leaderboard | Same |
| | 59.6% MMMU (val), IT (Table 6 and Table 16) | 59.6% Leaderboard | Same |
| | 48.8% MMMU (val), IT (Table 6 and Table 16) | 48.8% Leaderboard | Same |
One logged change that moves scores
Changes by the maintainers, so a score from before a change may not compare with one from after it. Newest first, each with the maintainers' own source. The list may be incomplete. What we log.
- Leaderboard rules changed February 12th, 2026
The MMMU team released the test answers, allowing local test scoring. Later test results no longer come exclusively from evaluation against hidden answers. Source
Versions
More in Vision
- MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
- CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
- MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy.
Seed2.1 Pro 90.7%
- MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy.
Qwen 3.8 Max 97.7%
- MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy.
Qwen 3.8 Max 85.9%
- ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance.
Command A Vision 90.9%
Data audit
All 10 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (186 entries), launch posts (93 scores) and system cards (51 scores).
- Pass Models scored 202 models we track, plus 49 other entries on its leaderboard.
- Pass Newest score The newest is from October 2026, for MiniCPM-V-4.7-35B-A3B.
- Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is Claude Opus 4.5 (#62).
- Note Headroom Saturated. The best is 86%, past 85% since November 2025.
Leaderboard import 3 pass
- Pass Importer We read the board on mmmu-benchmark.github.io: 140 rows for models we track and 49 other entries at the last read.
- Pass Last read November 18th, 2026.
- Pass Read worked The last read worked.
Consistency 2 pass
- Pass Lab claims All 39 pairs agree within 10 points.
- Pass Known issues No published audit found problems.