Benchmark
多模态开源模型榜
vision
multimodal
多语言
多模态开源模型榜：多维度评测综合成绩。评测分数为事实数据，整理自公开评测体系，仅供学习研究参考。
Top 10
榜单概览
数据说明
分数为 0-100 百分制综合成绩（Elo 榜为相对评分，仅供同榜内比较）。评测分数属事实数据，整理自公开评测体系与厂商公开资料，部分分数为厂商自报口径，不同榜单的评测方法与数据集可能存在差异，请结合多个榜单交叉参考。
Ranking
模型排名
| 名次 | 模型 | 机构 | 主分 | MMBench TEST | MMStar | MMMU VAL | MathVista | HallusionBench | AI2D | OCRBench | MMVet |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | JT3.5 | China Mobile | 82.7 | 88.3 | 83.8 | 80.1 | 82.7 | 63.8 | 88.7 | 91.5 | 82.9 |
| 2 | SenseNova-V6-5-Pro | SenseTime | 82.2 | 87.3 | 76.1 | 77.0 | 82.8 | 66.7 | 90.2 | 88.5 | 89.4 |
| 3 | TeleMM-2.0-Thinking | China Telecom | 81.1 | 87.9 | 73.7 | 75.8 | 82.8 | 65.2 | 89.5 | 90.8 | 82.8 |
| 4 | CongRong-v2.0 | CloudWalk | 80.7 | 88.1 | 75.3 | 75.6 | 76.8 | 63.2 | 90.0 | 92.7 | 83.9 |
| 5 | Gemini 2.5 Pro | 80.1 | 88.3 | 73.6 | 74.7 | 80.9 | 64.1 | 89.5 | 86.2 | 83.3 | |
| 6 | TeleMM-2.0 | China Telecom | 79.9 | 87.3 | 70.7 | 76.0 | 81.2 | 65.7 | 86.5 | 90.6 | 81.5 |
| 7 | GPT-5 | OpenAI | 79.9 | 86.6 | 75.7 | 81.8 | 81.9 | 65.2 | 89.5 | 80.7 | 77.6 |
| 8 | InternVL3-78B | Shanghai AI Laboratory & Tsinghua University | 79.1 | 87.7 | 73.4 | 72.2 | 79.0 | 59.1 | 89.8 | 90.8 | 80.7 |
| 9 | RBDashMM3-DPO-38B | RBDash Team | 78.9 | 87.2 | 72.1 | 70.9 | 79.3 | 61.6 | 88.7 | 90.1 | 81.2 |
| 10 | MiniCPM-o-4_5 | OpenBMB | 78.5 | 86.2 | 74.5 | 72.4 | 82.2 | 65.0 | 88.1 | 87.9 | 71.7 |
| 11 | BlueLM-2.5-3B-0909 | Vivo | 78.4 | 86.4 | 80.1 | 62.4 | 82.3 | 63.1 | 86.1 | 88.1 | 78.5 |
| 12 | GPT-5 mini | OpenAI | 78.0 | 86.2 | 73.2 | 78.7 | 79.2 | 62.5 | 86.7 | 82.8 | 74.6 |
| 13 | Ovis2.5-9B | AIDC-AI | 77.8 | 85.3 | 72.5 | 70.1 | 82.1 | 65.3 | 87.4 | 86.1 | 73.5 |
| 14 | InternVL3-38B | Shanghai AI Laboratory & Tsinghua University | 77.8 | 86.8 | 72.6 | 69.7 | 76.3 | 58.4 | 88.7 | 88.6 | 81.1 |
| 15 | Step-1o | StepFun | 77.7 | 87.3 | 69.3 | 69.9 | 74.7 | 55.8 | 89.1 | 92.6 | 82.8 |
| 16 | MiniCPM-V-4_5 | OpenBMB | 76.9 | 84.3 | 71.7 | 65.2 | 80.2 | 61.3 | 86.5 | 89.1 | 76.9 |
| 17 | GLM-4v-Plus-20250111 | Zhipu AI | 76.7 | 85.9 | 72.5 | 69.9 | 73.5 | 58.5 | 86.7 | 90.8 | 75.7 |
| 18 | Ovis2-34B | Alibaba Cloud / Qwen Team | 76.5 | 86.5 | 69.2 | 66.7 | 76.1 | 58.8 | 88.3 | 89.4 | 77.1 |
| 19 | Doubao-1.5-Pro | ByteDance | 76.5 | 82.1 | 71.9 | 71.4 | 78.6 | 58.5 | 87.0 | 82.9 | 79.8 |
| 20 | HunYuan-Standard-Vision | Tencent | 76.3 | 83.6 | 72.9 | 70.7 | 66.1 | 57.7 | 93.2 | 86.1 | 79.9 |
| 21 | Qwen2.5-VL-72B | Alibaba Cloud / Qwen Team | 76.1 | 87.8 | 70.5 | 68.2 | 74.2 | 54.6 | 88.5 | 88.2 | 76.9 |
| 22 | GPT-4.1 | OpenAI | 75.9 | 86.6 | 69.8 | 74.0 | 70.4 | 58.5 | 85.9 | 83.4 | 78.8 |
| 23 | R-4B | Unknown | 75.5 | 82.8 | 72.6 | 64.7 | 78.0 | 60.0 | 86.2 | 83.7 | 76.2 |
| 24 | ChatGPT-4o-latest | OpenAI | 75.4 | 86.0 | 70.2 | 72.9 | 71.6 | 57.0 | 86.3 | 82.2 | 76.9 |
| 25 | GPT-4.5 | OpenAI | 75.3 | 83.4 | 69.3 | 72.1 | 70.5 | 60.0 | 87.2 | 84.5 | 75.3 |
| 26 | BailingMM-Pro-0120 | Ant Group | 75.2 | 87.1 | 69.4 | 59.2 | 72.3 | 59.4 | 87.2 | 89.4 | 77.6 |
| 27 | JT-VL-Chat-Mini | China Mobile | 75.2 | 85.2 | 75.6 | 57.2 | 66.7 | 64.1 | 87.0 | 94.2 | 71.6 |
| 28 | InternVL3-14B | Shanghai AI Laboratory & Tsinghua University | 75.2 | 83.6 | 68.9 | 64.8 | 74.4 | 55.9 | 86.0 | 87.7 | 80.5 |
| 29 | Qwen2.5-VL-32B | Alibaba Cloud / Qwen Team | 74.8 | 84.0 | 70.3 | 68.9 | 72.8 | 58.4 | 85.4 | 85.6 | 72.9 |
| 30 | Gemini 2.0 Flash | 74.3 | 85.2 | 69.4 | 69.9 | 70.4 | 58.0 | 83.1 | 85.1 | 73.6 | |
| 31 | Kimi-VL-A3B-Thinking-2506 | Moonshot AI | 74.3 | 80.1 | 70.0 | 62.9 | 79.5 | 59.8 | 83.1 | 82.3 | 77.0 |
| 32 | SAIL-VL2-8B | ByteDance | 74.0 | 84.5 | 70.7 | 57.6 | 76.2 | 53.7 | 87.8 | 91.2 | 70.7 |
| 33 | SAIL-VL-1.6-8B | ByteDance | 73.6 | 84.0 | 69.5 | 55.4 | 74.2 | 54.4 | 87.5 | 90.5 | 73.3 |
| 34 | InternVL3-8B | Shanghai AI Laboratory & Tsinghua University | 73.6 | 82.1 | 68.7 | 62.2 | 70.5 | 49.0 | 85.1 | 88.4 | 82.8 |
| 35 | Ovis2-16B | Alibaba Cloud / Qwen Team | 73.3 | 85.7 | 67.2 | 60.7 | 73.7 | 56.8 | 86.3 | 87.9 | 68.4 |
| 36 | Gemini-2.0-Pro | 73.3 | 83.0 | 68.5 | 72.6 | 71.3 | 49.8 | 84.8 | 86.3 | 70.4 | |
| 37 | Valley3 | ByteDance | 73.1 | 83.1 | 66.7 | 63.2 | 75.2 | 58.1 | 84.6 | 87.2 | 66.5 |
| 38 | Ovis2.5-2B | AIDC-AI | 73.0 | 79.2 | 67.7 | 58.7 | 79.8 | 58.5 | 85.0 | 86.8 | 68.3 |
| 39 | JT-VL-Chat-2B | China Mobile | 72.7 | 82.5 | 74.3 | 54.9 | 63.0 | 61.7 | 84.5 | 94.5 | 65.9 |
| 40 | VARCO-VISION-2.0-14B | NCSOFT | 72.6 | 83.7 | 66.9 | 61.9 | 73.2 | 53.2 | 85.7 | 86.9 | 68.9 |
| 41 | InternVL3-9B | Shanghai AI Laboratory & Tsinghua University | 72.6 | 82.2 | 67.4 | 59.4 | 69.0 | 50.8 | 85.2 | 88.1 | 78.4 |
| 42 | Ola-7b | THU | 72.6 | 84.3 | 70.8 | 57.0 | 68.4 | 53.5 | 86.1 | 82.2 | 78.6 |
| 43 | WeThink-Qwen2.5VL-7B | WeChat Vision | 72.5 | 81.8 | 64.2 | 61.0 | 71.6 | 55.3 | 84.0 | 89.0 | 73.2 |
| 44 | SAIL-VL-1.5-8B | ByteDance | 72.4 | 82.1 | 69.7 | 54.3 | 73.4 | 54.5 | 87.5 | 88.9 | 69.0 |
| 45 | GPT-5 nano | OpenAI | 72.2 | 79.2 | 68.3 | 72.6 | 73.1 | 60.9 | 81.4 | 74.7 | 67.4 |
| 46 | Ovis2-8B | Alibaba Cloud / Qwen Team | 71.8 | 83.6 | 64.6 | 57.4 | 71.8 | 56.3 | 86.6 | 89.1 | 65.1 |
| 47 | MUG-U-7B | Shopee | 71.5 | 81.8 | 66.6 | 54.3 | 74.8 | 51.3 | 88.9 | 91.1 | 63.0 |
| 48 | Qwen2.5-VL-7B | Alibaba Cloud / Qwen Team | 70.9 | 82.2 | 64.1 | 58.0 | 68.1 | 51.9 | 84.3 | 88.8 | 69.7 |
| 49 | BlueLM-2.5-3B | Vivo | 70.8 | 74.6 | 67.8 | 58.7 | 76.7 | 60.0 | 83.1 | 83.8 | 61.7 |
| 50 | MiniCPM-o-2.6 | OpenBMB | 70.2 | 80.6 | 63.3 | 50.9 | 73.3 | 51.1 | 86.1 | 88.9 | 67.2 |
| 51 | Claude3.7-Sonnet | Anthropic | 70.1 | 79.7 | 65.1 | 71.0 | 66.8 | 55.4 | 82.5 | 70.1 | 70.0 |
| 52 | Ovis2-4B | Alibaba Cloud / Qwen Team | 69.8 | 81.4 | 61.9 | 49.0 | 69.6 | 53.8 | 85.7 | 91.1 | 65.5 |
| 53 | SAIL-VL2-2B | Douyin | 69.4 | 80.1 | 64.0 | 49.3 | 71.5 | 51.1 | 83.1 | 90.6 | 65.6 |
| 54 | MiniCPM-V-4 | OpenBMB | 69.4 | 79.7 | 61.9 | 54.0 | 67.5 | 52.6 | 82.9 | 89.7 | 66.5 |
| 55 | Ovis-U1-3B | Alibaba Cloud / Qwen Team | 69.3 | 77.9 | 61.3 | 50.6 | 68.9 | 55.8 | 85.6 | 88.1 | 66.0 |
| 56 | valley2_dpo | ByteDance | 69.3 | 81.8 | 61.9 | 58.4 | 68.7 | 52.4 | 83.0 | 85.8 | 62.3 |
| 57 | Kimi-VL-A3B-Instruct | Moonshot AI | 69.1 | 80.8 | 62.0 | 57.8 | 66.0 | 48.4 | 84.5 | 87.1 | 66.1 |
| 58 | GPT-4.1 mini | OpenAI | 68.9 | 80.9 | 60.9 | 55.0 | 70.9 | 49.3 | 76.0 | 84.0 | 74.3 |
| 59 | MiniCPM-V-4.6 | OpenBMB | 68.1 | 77.3 | 65.7 | 51.9 | 74.4 | 53.9 | 82.4 | 80.6 | 58.6 |
| 60 | Ristretto-3B | LiAutoAD | 67.7 | 80.2 | 62.8 | 51.3 | 67.6 | 50.2 | 84.2 | 84.7 | 60.7 |
| 61 | Gemma3-27B | 67.4 | 78.9 | 59.6 | 64.8 | 59.8 | 48.8 | 81.4 | 75.3 | 71.0 | |
| 62 | valley2 | ByteDance | 67.1 | 81.0 | 60.0 | 56.8 | 61.7 | 47.6 | 82.8 | 84.5 | 62.1 |
| 63 | SAIL-VL-1.5-2B | ByteDance | 67.0 | 78.5 | 62.6 | 46.4 | 67.0 | 50.0 | 83.7 | 89.1 | 58.8 |
| 64 | grok-2-vision-1212 | xAI | 66.9 | 82.9 | 65.2 | 67.1 | 66.6 | 51.7 | 81.2 | 55.5 | 65.2 |
| 65 | VARCO-VISION-14B | NC Research | 66.8 | 81.2 | 64.1 | 56.3 | 67.6 | 46.8 | 83.9 | 81.5 | 53.0 |
| 66 | DeepSeek VL2 | DeepSeek | 66.4 | 81.2 | 61.9 | 54.0 | 63.9 | 45.3 | 83.8 | 80.9 | 60.0 |
| 67 | QTuneVL1.5-3B | Reconova AI Lab & BDAA-Lab | 66.1 | 77.3 | 57.1 | 53.4 | 64.4 | 49.0 | 81.2 | 83.9 | 62.8 |
| 68 | Kimi-VL-A3B-Thinking | Moonshot AI | 65.9 | 68.3 | 60.8 | 50.7 | 67.6 | 52.3 | 77.5 | 82.5 | 67.4 |
| 69 | MiMo-VL-7B | Xiaomi | 65.6 | 80.7 | 38.9 | 26.4 | 77.4 | 63.8 | 82.8 | 82.9 | 72.2 |
| 70 | Ovis2-2B | Alibaba Cloud / Qwen Team | 65.2 | 76.9 | 56.7 | 45.6 | 64.1 | 50.2 | 82.7 | 87.3 | 58.3 |
| 71 | Phi-4-MultiModal | Microsoft | 64.7 | 77.2 | 58.9 | 56.0 | 65.8 | 40.5 | 83.0 | 84.4 | 51.9 |
| 72 | QTuneVL1.5-2B | Reconova AI Lab & BDAA-Lab | 64.6 | 77.8 | 60.8 | 48.7 | 56.1 | 43.4 | 78.6 | 85.8 | 65.9 |
| 73 | Qwen2.5-VL-3B | Alibaba Cloud / Qwen Team | 64.5 | 76.8 | 56.3 | 51.2 | 61.2 | 46.6 | 81.4 | 82.8 | 60.0 |
| 74 | DeepSeek VL2 Small | DeepSeek | 64.5 | 79.9 | 57.5 | 49.7 | 61.9 | 43.8 | 81.7 | 83.2 | 58.6 |
| 75 | InternVL3-2B | Shanghai AI Laboratory & Tsinghua University | 64.5 | 78.0 | 61.1 | 48.7 | 57.6 | 41.9 | 78.6 | 83.1 | 67.0 |
| 76 | VITA-1.5 | NJU | 63.3 | 76.8 | 60.2 | 52.6 | 66.2 | 44.6 | 79.2 | 74.1 | 52.7 |
| 77 | Gemma3-12B | 62.8 | 74.6 | 56.1 | 55.2 | 56.1 | 47.2 | 78.1 | 70.2 | 64.9 | |
| 78 | LLaVA-CoT | Peking University | 61.9 | 72.9 | 54.9 | 53.7 | 52.5 | 47.7 | 78.7 | 75.5 | 59.5 |
| 79 | Flash-VL-2B-Dynamic-ISS | Meituan | 61.6 | 75.8 | 53.8 | 45.6 | 61.7 | 48.5 | 74.4 | 84.4 | 48.6 |
| 80 | Long-VITA-16K | Tencent | 61.1 | 76.3 | 60.1 | 53.9 | 64.8 | 45.5 | 80.9 | 76.1 | 31.0 |
| 81 | QTuneVL1-2B | Reconova AI Lab | 59.9 | 71.1 | 53.6 | 44.7 | 48.8 | 42.4 | 75.2 | 80.6 | 63.0 |
| 82 | Ovis2-1B | Alibaba Cloud / Qwen Team | 59.6 | 68.3 | 52.1 | 36.1 | 59.4 | 45.2 | 76.4 | 89.0 | 50.0 |
| 83 | VARCO-VISION-2.0-1.7B | NCSOFT | 59.2 | 72.8 | 52.5 | 44.2 | 55.8 | 38.6 | 75.2 | 80.8 | 53.4 |
| 84 | granite-vision-3.3-2b | IBM | 58.7 | 68.2 | 51.9 | 39.0 | 60.2 | 47.0 | 76.6 | 82.0 | 45.0 |
| 85 | DeepSeek VL2 Tiny | DeepSeek | 58.1 | 70.9 | 49.8 | 42.9 | 54.0 | 39.6 | 74.6 | 80.5 | 52.5 |
| 86 | granite-vision-3.2-2b | IBM | 57.9 | 70.8 | 54.6 | 39.0 | 59.7 | 40.7 | 78.6 | 79.0 | 40.5 |
| 87 | GPT-4.1 nano | OpenAI | 57.5 | 62.4 | 47.5 | 57.6 | 53.7 | 36.9 | 68.0 | 71.0 | 62.8 |
| 88 | granite-vision-3.1-2b-preview | IBM | 57.3 | 69.2 | 56.1 | 35.4 | 60.5 | 41.4 | 77.7 | 78.3 | 39.6 |
| 89 | InternVL3-1B | Shanghai AI Laboratory & Tsinghua University | 57.0 | 68.2 | 52.3 | 43.2 | 46.9 | 37.2 | 69.7 | 79.8 | 58.7 |
| 90 | Ross-Qwen2-7B | CASIA | 55.9 | 75.4 | 53.9 | 49.0 | 52.5 | 38.7 | 79.4 | 55.3 | 43.2 |
| 91 | Gemma3-4B | 55.4 | 66.4 | 47.9 | 47.3 | 46.3 | 40.8 | 70.7 | 66.0 | 57.8 | |
| 92 | VITA | NJU | 54.2 | 69.8 | 46.6 | 47.6 | 44.5 | 41.0 | 72.8 | 68.3 | 43.2 |
| 93 | HawkVL-2B | Reconova AI Lab & USTC | 52.8 | 64.5 | 48.1 | 43.6 | 43.3 | 40.9 | 66.6 | 76.3 | 38.9 |
| 94 | SmolVLM2 | HuggingFace | 52.2 | 61.1 | 46.0 | 41.6 | 51.5 | 40.6 | 69.7 | 72.5 | 34.9 |
| 95 | Janus-Pro-7B | DeepSeek | 50.2 | 59.1 | 46.5 | 41.6 | 42.5 | 39.5 | 68.1 | 59.5 | 45.1 |
| 96 | SmolVLM-Instruct | HuggingFace | 48.1 | 57.4 | 41.7 | 38.8 | 43.6 | 39.5 | 64.2 | 65.6 | 33.8 |
| 97 | Emu3_chat | Beijing Academy of Artificial Intelligence | 47.5 | 52.7 | 46.6 | 33.9 | 47.6 | 31.7 | 71.5 | 66.7 | 29.1 |
| 98 | SmolVLM-Instruct-DPO | HuggingFace | 44.8 | 55.4 | 39.2 | 39.7 | 42.9 | 39.8 | 62.2 | 50.0 | 29.3 |
| 99 | SmolVLM-500M | HuggingFace | 41.3 | 41.9 | 38.3 | 33.6 | 39.8 | 31.1 | 59.2 | 60.9 | 25.7 |
| 100 | SmolVLM2-500M | HuggingFace | 40.9 | 41.6 | 38.2 | 34.1 | 37.5 | 27.7 | 57.3 | 60.9 | 29.9 |
| 101 | AKI-4B | Sony | 39.4 | 55.6 | 37.8 | 36.3 | 28.5 | 33.9 | 56.5 | 30.8 | 36.2 |
| 102 | SmolVLM-Synthetic | HuggingFace | 38.6 | 43.1 | 34.9 | 38.6 | 31.5 | 34.8 | 50.8 | 46.1 | 29.3 |
| 103 | SmolVLM-256M | HuggingFace | 34.4 | 26.9 | 34.5 | 28.9 | 37.0 | 27.9 | 46.4 | 52.7 | 21.1 |
| 104 | SmolVLM2-256M | HuggingFace | 30.8 | 6.5 | 30.9 | 29.0 | 34.0 | 24.3 | 39.2 | 56.6 | 25.5 |