벤치마크 상세
MMMU
이미지와 전문 지식을 함께 이해: 차트, 문서 이미지, 시각 문제, 이미지 기반 질의응답
공식 리더보드 보기
Validation의 Overall만 비교합니다. Test와 MMMU-Pro 점수를 섞지 않고 사람 기준선은 모델 순위에서 제외했습니다.
모든 공개 결과
모델 검색 · 점수순확인 2026. 09. 07.MMMU validation · Overall% validation accuracy
189개 결과원문 ↗
| 순서 | 모델 / 실행 구성 | 점수 | 평가 조건·근거 |
|---|---|---|---|
| 1 | GPT-5.1 | 85.4 | 2025-11-13 · author |
| 2 | DreamPRM-1.5 (GPT-5-mini w/ thinking) | 84.6 | 2025-09-05 · author |
| 3 | GPT-5 w/ thinking | 84.2 | 2025-08-07 · author |
| 4 | Gemini 2.5 Pro Deep-Think | 84 | 2025-05-20 · author |
| 5 | o3 | 82.9 | 2025-04-16 · author |
| 6 | o4-mini | 81.6 | 2025-04-16 · author |
| 7 | Claude Opus 4.5 | 80.7 | 2025-11-24 · author |
| 8 | dots.vlm1 | 80.1 | 2025-08-05 · author |
| 9 | Gemini 2.5 Flash 05-20 | 79.7 | 2025-05-20 · author |
| 10 | Gemini 2.5 Pro 05-06 | 79.6 | 2025-05-06 · author |
| 11 | Qwen3-VL 235B-A22B | 78.7 | 2025-09-22 · author |
| 12 | o1 | 78.2 | 2024-09-12 · author |
| 13 | Grok 3 Beta | 78 | 2025-02-19 · author |
| 14 | Seed 1.5-VL Thinking | 77.9 | 2025-05-13 · author |
| 15 | Claude Sonnet 4.5 | 77.8 | 2025-09-29 · author |
| 16 | Intern-S1 | 77.7 | 2025-08-04 · author |
| 17 | Claude Opus 4.1 | 77.1 | 2025-08-05 · author |
| 18 | Claude Opus 4 | 76.5 | 2025-05-23 · author |
| 19 | Llama 4 Behemoth | 76.1 | 2025-04-05 · author |
| 20 | Skywork-R1V3-38B | 76 | 2025-07-05 · author |
| 21 | GLM-4.5V w/ Thinking | 75.4 | 2025-08-11 · author |
| 22 | Claude 3.7 Sonnet | 75 | 2025-02-24 · author |
| 23 | Seed 1.6-Thinking | 74.8 | 2025-06-25 · author |
| 24 | GPT-5 w/o thinking | 74.4 | 2025-08-07 · author |
| 25 | GPT‑4.5 | 74.4 | 2025-02-28 · author |
| 26 | Claude Sonnet 4 | 74.4 | 2025-05-23 · author |
| 27 | Seed 1.5-VL | 73.6 | 2025-05-13 · author |
| 28 | Skywork-R1V2-38B | 73.6 | 2025-04-28 · author |
| 29 | Llama 4 Maverick | 73.4 | 2025-04-05 · author |
| 30 | Gemini 2.0 Pro | 72.7 | 2025-02-05 · author |
| 31 | KiQ-v0 | 72.6 | 2025-08-31 · official leaderboard |
| 32 | InternVL3-78B | 72.2 | 2025-04-11 · author |
| 33 | LLaVA-Critic-R1+ | 71.9 | 2025-08-30 · author |
| 34 | Gemini 2.0 Flash | 71.7 | 2025-02-05 · author |
| 35 | GPT-4o (1120) | 70.7 | 2024-11-20 · author |
| 36 | MiMo-VL 7B-RL-2508 w/ Thinking | 70.6 | 2025-08-08 · author |
| 37 | Spark4.0-Turbo | 70.4 | 2025-01-22 · author |
| 38 | QVQ-72B-Preview | 70.3 | 2024-12-25 · author |
| 39 | InternVL2.5-78B | 70.1 | 2024-12-05 · author |
| 40 | InternVL3-38B | 70.1 | 2025-04-11 · author |
| 41 | LLaVA-Critic-R1 | 69.7 | 2025-08-30 · author |
| 42 | Llama 4 Scout | 69.4 | 2025-04-05 · author |
| 43 | GPT-4o (0513) | 69.1 | 2024-05-13 · author |
| 44 | Skywork-R1V-38B | 69 | 2025-04-28 · author |
| 45 | Claude 3.5 Sonnet | 68.3 | 2024-06-20 · official leaderboard |
| 46 | Gemini 2.0 Flash-Lite | 68 | 2025-02-05 · author |
| 47 | GLM-4.1V w/ Thinking | 68 | 2025-07-01 · author |
| 48 | Nova Premier | 68 | 2025-04-30 · author |
| 49 | InternVL3-14B | 67.1 | 2025-04-11 · author |
| 50 | MiMo-VL 7B-RL | 66.7 | 2025-06-04 · author |
| 51 | Gemini 1.5 Pro (0801) | 65.8 | 2024-08-01 · official leaderboard |
| 52 | Ovis2-34B | 65.6 | 2025-03-25 · author |
| 53 | Gemma 3-27B | 64.9 | 2025-03-12 · author |
| 54 | MiMo-VL 7B-SFT | 64.6 | 2025-06-04 · author |
| 55 | Qwen2-VL-72B | 64.5 | 2024-08-29 · official leaderboard |
| 56 | Qwen2.5-VL 72B | 64.5 | 2025-02-19 · author |
| 57 | InternVL2.5-38B | 63.9 | 2024-12-05 · author |
| 58 | MiMo-VL 7B-RL-2508 w/o Thinking | 63.1 | 2025-08-08 · author |
| 59 | InternVL3-8B | 62.7 | 2025-04-11 · author |
| 60 | Gemini 1.5 Pro (0523) | 62.2 | 2024-05-23 · author |
| 61 | InternVL2-Pro | 62 | 2024-07-04 · author |
| 62 | Nova Pro | 62 | 2025-04-30 · author |
| 63 | TeleMM | 61.4 | 2024-11-18 · author |
| 64 | Llama 3.2 90B | 60.3 | 2024-09-25 · author |
| 65 | NVLM-H 1.0 78B | 60.2 | 2024-09-17 · author |
| 66 | InternVL2.5-26B | 60 | 2024-12-05 · author |
| 67 | NVLM-D 1.0 78B | 59.7 | 2024-09-17 · author |
| 68 | Gemma 3-12B | 59.6 | 2025-03-12 · author |
| 69 | Ovis2-16B | 59.6 | 2025-03-25 · author |
| 70 | Gemini 1.0 Ultra | 59.4 | 2023-12-06 · author |
| 71 | Claude 3 Opus | 59.4 | 2024-03-05 · author |
| 72 | GPT-4o mini | 59.4 | 2024-07-18 · author |
| 73 | Ovis2-8B | 59 | 2025-03-25 · author |
| 74 | NVILA | 58.7 | 2024-11-20 · author |
| 75 | Qwen2.5-VL 7B | 58.6 | 2025-02-19 · author |
| 76 | Qianfan-VL-70B | 58.3 | 2025-09-16 · author |
| 77 | InternVL2-Llama3-76B | 58.3 | 2024-07-04 · author |
| 78 | RBDash V1.2 | 57.7 | 2024-08-21 · author |
| 79 | InternVL3-9B | 57.7 | 2025-04-11 · author |
| 80 | Kimi-VL-A3B | 57 | 2025-04-10 · author |
| 81 | GPT-4V(ision) (Playground) | 56.8 | 2023-11-27 · official leaderboard |
| 82 | LLaVA-OneVision-72B | 56.8 | 2024-08-06 · author |
| 83 | Reka Core | 56.3 | 2024-04-23 · author |
| 84 | Gemini 1.5 Flash | 56.1 | 2024-05-23 · author |
| 85 | InternVL2.5-8B | 56 | 2024-12-05 · author |
| 86 | Eagle2.5-8B | 55.8 | 2025-04-21 · author |
| 87 | InternVL2-40B | 55.2 | 2024-07-04 · author |
| 88 | Aria | 54.9 | 2024-10-10 · author |
| 89 | SenseChat-Vision-0423-Preview | 54.6 | 2024-04-23 · author |
| 90 | POINTS1.5 | 53.8 | 2024-12-11 · author |
| 91 | Reka Flash | 53.3 | 2024-04-23 · author |
| 92 | Claude 3 Sonnet | 53.1 | 2024-03-05 · author |
| 93 | Qwen2.5-VL 3B | 53.1 | 2025-02-19 · author |
| 94 | EVLM-KTO | 53 | 2024-12-13 · author |
| 95 | HPT Pro | 52 | 2024-03-16 · author |
| 96 | InternVL-Chat-V1.2 | 51.6 | 2024-02-22 · author |
| 97 | Qwen-VL-MAX | 51.4 | 2024-01-27 · author |
| 98 | Skywork-VL | 51.4 | 2024-05-30 · author |
| 99 | InternVL2-8B | 51.2 | 2024-07-04 · author |
| 100 | DeepSeek-VL2 | 51.1 | 2024-12-13 · author |
| 101 | SpaRE-7B | 51 | 2025-04-29 · author |
| 102 | 360VL | 50.8 | 2024-05-21 · author |
| 103 | MAmmoTH-VL-8B | 50.8 | 2024-12-08 · author |
| 104 | Llama 3.2 11B | 50.7 | 2024-09-25 · author |
| 105 | BLIP3-o8B | 50.6 | 2025-05-16 · author |
| 106 | Claude 3 Haiku | 50.2 | 2024-03-05 · author |
| 107 | LlaVA-NEXT-72B | 49.9 | 2024-01-30 · author |
| 108 | MiniCPM-V 2.6 | 49.8 | 2024-08-06 · author |
| 109 | Ovis1.5-Gemma2-9B | 49.8 | 2024-06-14 · author |
| 110 | Cambrian-1-34B | 49.7 | 2024-06-27 · author |
| 111 | LLaVA-OneVision-7B | 48.8 | 2024-08-06 · author |
| 112 | Gemma 3-4B | 48.8 | 2025-03-12 · author |
| 113 | Ovis-Clip-Llama3-8B | 48.6 | 2024-06-14 · author |
| 114 | InternVL3-2B | 48.6 | 2025-04-11 · author |
| 115 | Adept Fuyu-Heavy | 48.3 | 2024-01-31 · author |
| 116 | LLaVA-NEXT-34B | 48.1 | 2024-01-30 · author |
| 117 | DeepSeek-VL2-Small | 48 | 2024-12-13 · author |
| 118 | Ovis2-4B | 48 | 2025-03-25 · author |
| 119 | Gemini 1.0 Pro | 47.9 | 2023-12-06 · author |
| 120 | Aquila-VL-2B | 47.4 | 2024-10-25 · author |
| 121 | Qianfan-VL-8B | 47.1 | 2025-09-16 · author |
| 122 | Idefics3-8B-Llama3 | 46.6 | 2024-08-22 · author |
| 123 | BLIP3-o4B | 46.6 | 2025-05-16 · author |
| 124 | Qianfan-VL-3B | 46.4 | 2025-09-16 · author |
| 125 | Marco-VL-Plus | 46.2 | 2024-03-09 · author |
| 126 | Yi-VL-34B | 45.9 | 2023-12-31 · author |
| 127 | MiniCPM-Llama3-V 2.5 | 45.8 | 2024-05-20 · author |
| 128 | Ovis2-2B | 45.3 | 2025-03-25 · author |
| 129 | Qwen-VL-PLUS | 45.2 | 2023-12-15 · author |
| 130 | Aya-Vision-32B | 45.11 | 2025-05-13 · author |
| 131 | Mogao-7B | 44.2 | 2025-05-08 · author |
| 132 | HPT Air | 44 | 2024-03-16 · author |
| 133 | InternVL2.5-2B | 43.6 | 2024-12-05 · author |
| 134 | InternVL3-1B | 43.4 | 2025-04-11 · author |
| 135 | Bunny-8B | 43.3 | 2024-06-25 · author |
| 136 | InternLM-XComposer2-VL | 43 | 2024-01-22 · author |
| 137 | Phi-3.5-Vision | 43 | 2024-05-21 · author |
| 138 | Flash-VL-2B-Dynamic | 42.9 | 2025-05-16 · author |
| 139 | Reka Edge | 42.8 | 2024-04-23 · author |
| 140 | Cambrian-1-8B | 42.7 | 2024-06-27 · author |
| 141 | MetaMorph | 41.8 | 2024-12-19 · author |
| 142 | HyViLM | 41.8 | 2024-12-11 · author |
| 143 | Bunny-4B | 41.4 | 2024-06-25 · author |
| 144 | Marco-VL | 41.2 | 2024-01-16 · author |
| 145 | OmniLMM-12B | 41.1 | 2024-02-07 · author |
| 146 | InternVL2.5-1B | 40.9 | 2024-12-05 · author |
| 147 | DeepSeek-VL2-Tiny | 40.7 | 2024-12-13 · author |
| 148 | Cambrian-1-13B | 40 | 2024-06-27 · author |
| 149 | SpaRE-2B | 40 | 2025-04-29 · author |
| 150 | Aya-Vision-8B | 39.9 | 2025-05-13 · author |
| 151 | InfiMM-Zephyr-7B | 39.4 | 2024-01-18 · author |
| 152 | Yi-VL-6B | 39.1 | 2024-01-23 · author |
| 153 | InternVL-Chat-V1.1 | 39.1 | 2024-01-24 · author |
| 154 | Math-LLaVA-13B | 38.3 | 2024-06-25 · author |
| 155 | Bunny-3B | 38.2 | 2024-02-13 · author |
| 156 | LLaVA-UHD v2 | 38.2 | 2024-12-21 · author |
| 157 | ILLUME | 38.2 | 2024-12-09 · author |
| 158 | SVIT | 38 | 2023-12-26 · author |
| 159 | MiniCPM-V | 37.2 | 2024-02-01 · author |
| 160 | MiniCPM-V 2.0 | 37.1 | 2024-04-12 · author |
| 161 | LLaVA-1.5-13B | 36.4 | 2023-11-27 · official leaderboard |
| 162 | Emu2-Chat | 36.3 | 2023-12-24 · author |
| 163 | LLaVA-NeXT-13B | 36.2 | 2024-01-30 · author |
| 164 | Ovis2-1B | 36 | 2025-03-25 · author |
| 165 | Qwen-VL-7B-Chat | 35.9 | 2023-11-27 · official leaderboard |
| 166 | LLaVA-NeXT-Vicuna-7B | 35.8 | 2024-01-30 · author |
| 167 | InstructBLIP-T5-XXL | 35.7 | 2023-11-27 · official leaderboard |
| 168 | BLIP-2 FLAN-T5-XXL | 35.4 | 2023-11-27 · official leaderboard |
| 169 | LLaVA-NeXT-mistral-7B | 35.3 | 2024-01-30 · author |
| 170 | BLIP-2 FLAN-T5-XL | 34.4 | 2023-11-27 · official leaderboard |
| 171 | SynerGen-VL | 34.2 | 2024-12-12 · author |
| 172 | HoVLE | 33.7 | 2024-12-18 · author |
| 173 | HoVLE (HD) | 33.2 | 2024-12-18 · author |
| 174 | InstructBLIP-T5-XL | 32.9 | 2023-11-27 · official leaderboard |
| 175 | SPHINX | 32.9 | 2023-12-26 · author |
| 176 | Olympus | 32.8 | 2024-12-12 · author |
| 177 | mPLUG-OWL2 | 32.7 | 2023-12-04 · author |
| 178 | Gemini Nano2 | 32.6 | 2023-12-06 · author |
| 179 | Otter | 32.2 | 2023-11-27 · official leaderboard |
| 180 | CogVLM | 32.1 | 2023-11-27 · official leaderboard |
| 181 | Emu3 | 31.6 | 2024-09-27 · author |
| 182 | LLaMA-Adapter2-7B | 29.8 | 2023-11-27 · official leaderboard |
| 183 | OpenFlamingo2-9B | 28.7 | 2023-11-27 · official leaderboard |
| 184 | Adept Fuyu-8B | 27.9 | 2023-11-27 · official leaderboard |
| 185 | Show-o | 27.4 | 2024-08-22 · author |
| 186 | MiniGPT4-Vicuna-13B | 26.8 | 2023-11-27 · official leaderboard |
| 187 | Frequent Choice | 26.8 | 2023-11-27 · official leaderboard |
| 188 | Kosmos2 | 24.4 | 2023-11-27 · official leaderboard |
| 189 | Random Choice | 22.1 | 2023-11-27 · official leaderboard |
일치하는 모델이 없습니다. 검색어를 줄여 보세요.
공개된 숫자만 높은 순으로 표시합니다. 순서가 신뢰구간을 고려한 확정 순위는 아닙니다. Validation의 Overall만 비교합니다. Test와 MMMU-Pro 점수를 섞지 않고 사람 기준선은 모델 순위에서 제외했습니다.
읽는 법
전체 벤치마크이미지와 텍스트를 함께 이해하는 multimodal reasoning을 평가합니다.
이미지를 예쁘게 만드는 능력이 아니라, 이미지를 읽고 추론하는 능력입니다.