Visual Question Answering on UniPercept-Bench-VQA IAA 1.0 (test)
80Comprehension ScoreUniPercept
Evaluation Results
| Method | Links | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| UniPerceptModel Category=Specialized, Training=Two-stage (Domain-Adaptive + Task-Aligned RL)2025.12 | 80 | 77.54 | 69.7 | 80.56 | 79.26 | 80.95 | 67.53 | 69.77 | 63.71 | 92.2 | 81.88 | 75.32 | 86.67 | 84.62 | 76.55 | |
| Gemini-2.5-proModel Category=Proprietary2025.12 | 71.79 | 68.45 | 61.59 | 76.85 | 67.41 | 63.81 | 61.84 | 37.21 | 45.98 | 78.72 | 73.91 | 67.72 | 84.67 | 84.62 | 66.44 | |
| InternVL3-78BModel Category=Open-Source, Parameters=78B2025.12 | 71.79 | 73.26 | 61.21 | 73.15 | 74.81 | 74.29 | 53.25 | 37.21 | 45.14 | 85.82 | 81.16 | 72.15 | 86 | 75.64 | 68.28 | |
| Claude-Sonnet-4.5-ThinkModel Category=Proprietary, Reasoning Mode=Enabled2025.12 | 71.28 | 69.52 | 61.21 | 68.52 | 62.22 | 66.67 | 53.25 | 41.86 | 44.57 | 75.89 | 77.54 | 67.09 | 86 | 66.67 | 64.73 | |
| Claude-Sonnet-4.5Model Category=Proprietary, Reasoning Mode=Disabled2025.12 | 70.26 | 70.05 | 62.2 | 71.3 | 64.44 | 67.62 | 50 | 46.51 | 46.84 | 77.3 | 76.09 | 65.19 | 86 | 69.23 | 65.45 | |
| ArtiMuseModel Category=Specialized2025.12 | 67.69 | 68.45 | 64.85 | 74.07 | 71.85 | 64.76 | 61.04 | 32.56 | 39.14 | 88.65 | 76.81 | 72.78 | 85.33 | 79.49 | 66.31 | |
| LLaVA-OneVision-1.5-Instruct-8BModel Category=Open-Source, Parameters=8B2025.12 | 67.18 | 68.62 | 61.21 | 62.96 | 67.41 | 62.86 | 53.25 | 20.93 | 34.86 | 85.21 | 79.71 | 65.82 | 83.33 | 69.23 | 62.6 | |
| GLM-4.5-V-106BA12BModel Category=Open-Source2025.12 | 67.18 | 65.78 | 60.98 | 75 | 64.44 | 68.57 | 51.32 | 46.51 | 45.4 | 71.63 | 78.26 | 65.82 | 84.67 | 70.51 | 64.46 | |
| QwenVL-2.5-Instruct-7BModel Category=Open-Source, Parameters=7B2025.12 | 67.18 | 70.74 | 56.36 | 66.67 | 68.89 | 63.81 | 48.05 | 37.21 | 38.86 | 76.76 | 75.36 | 67.09 | 87.33 | 71.79 | 63.19 | |
| InternVL3-8BModel Category=Open-Source, Parameters=8B2025.12 | 65.64 | 67.55 | 59.39 | 67.59 | 69.63 | 62.86 | 50.65 | 25.58 | 36 | 81.69 | 73.91 | 67.72 | 86 | 71.79 | 62.6 | |
| GPT-4oModel Category=Proprietary2025.12 | 64.62 | 59.57 | 57.58 | 60.19 | 65.19 | 67.62 | 51.95 | 30.23 | 38.86 | 78.17 | 72.46 | 62.66 | 72.67 | 70.51 | 60.04 | |
| Llama-4-ScoutModel Category=Proprietary2025.12 | 62.56 | 68.45 | 59.76 | 61.11 | 57.78 | 70.48 | 48.68 | 32.56 | 43.97 | 70.92 | 69.57 | 61.39 | 77.33 | 70.51 | 60.91 | |
| InternVL3.5-38BModel Category=Open-Source, Parameters=38B2025.12 | 37.44 | 40.11 | 27.88 | 39.81 | 34.81 | 38.1 | 45.45 | 6.98 | 34 | 47.52 | 26.09 | 28.48 | 37.33 | 50 | 35.67 | |
| InternVL3.5-8BModel Category=Open-Source, Parameters=8B2025.12 | 32.31 | 29.41 | 30.3 | 26.85 | 28.89 | 26.67 | 23.38 | 9.3 | 17.14 | 41.13 | 26.81 | 19.62 | 36 | 58.97 | 28.18 | |
| QwenVL-3-Instruct-8BModel Category=Open-Source, Parameters=8B2025.12 | 31.28 | 32.09 | 32.12 | 37.04 | 34.07 | 22.86 | 37.66 | 25.58 | 35.43 | 14.89 | 17.39 | 34.81 | 28.67 | 73.08 | 31.92 | |
| Random Guess2025.12 | 23.08 | 27.27 | 21.95 | 29.63 | 25.93 | 22.86 | 23.68 | 32.56 | 24.14 | 21.28 | 30.43 | 25.32 | 24 | 29.49 | 25.17 | |
| QwenVL-3-Instruct-32BModel Category=Open-Source, Parameters=32B2025.12 | 23.08 | 26.74 | 32.12 | 26.85 | 32.59 | 20.95 | 33.77 | 20.93 | 33.43 | 9.22 | 13.77 | 31.01 | 18.67 | 66.67 | 27.39 | |
| QwenVL-2.5-Instruct-72BModel Category=Open-Source, Parameters=72B2025.12 | 22.05 | 24.6 | 25.45 | 29.63 | 30.37 | 18.1 | 19.48 | 6.98 | 14 | 19.86 | 17.39 | 24.05 | 41.33 | 51.28 | 23.74 |