Visual Perception on BLINK (val)
95.67Validation ScoreHuman
Evaluation results
| Method | Links | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Human2026.02 | 95.67 | 80 | 96.7 | 93.75 | 99.19 | 99 | 95.3 | 80.77 | 96.07 | 98.25 | 98 | 99.42 | 92.48 | 95.14 | 100 | |
| Qwen3-VL-8BParameter Scale=8B2026.05 | 65.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Molmo2-4BParameter Scale=4B2026.05 | 63.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-4BParameter Scale=4B2026.05 | 63.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ReMoT -CoT-4BAccess=Open-source, Size=4B, Mode=Chain-of-Thought2026.02 | 62.15 | 23.2 | 84.44 | 73.33 | 80.65 | 59.33 | 75.21 | 46.92 | 64.03 | 86.71 | 53.28 | 87.21 | 42.86 | 28.36 | 56.06 | |
| GPT-4oAccess=API-based2026.02 | 60.04 | 31.33 | 72.59 | 49.17 | 74.19 | 55.33 | 82.91 | 40.77 | 53.96 | 69.23 | 59.84 | 75 | 59.4 | 37.31 | 79.55 | |
| InternVL3.5-8BParameter Scale=8B2026.05 | 59.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Molmo2-8BParameter Scale=8B2026.05 | 59.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3VL-4B-ThinkingAccess=Open-source, Size=4B, Mode=Thinking2026.02 | 59.5 | 10 | 61.54 | 71.67 | 80.65 | 58.67 | 77.42 | 39.23 | 56.12 | 88.11 | 52.46 | 86.63 | 49.62 | 33.58 | 60.61 | |
| InternVL3.5-4BParameter Scale=4B2026.05 | 58.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| V-ReflectionParam Size=7B2026.03 | 56.4 | 33.3 | — | 65.8 | — | 49.6 | — | — | — | 90.9 | — | — | — | 44.8 | — | |
| LVRParam Size=7B2026.03 | 55.4 | 29.3 | — | 70 | — | 52 | — | — | — | 86 | — | — | — | 42.5 | — | |
| PAPOParam Size=7B2026.03 | 54.8 | 29.3 | — | 66.7 | — | 52 | — | — | — | 88.8 | — | — | — | 39.6 | — | |
| GPT-4 TurboAccess=API-based2026.02 | 54.61 | 32.67 | 80.74 | 57.5 | 66.13 | 69.33 | 79.49 | 24.62 | 30.94 | 69.23 | 52.46 | 52.33 | 52.63 | 32.84 | 63.64 | |
| Qwen2.5-VL-7BParam Size=7B2026.03 | 54.5 | 27.3 | — | 65.8 | — | 52.7 | — | — | — | 88.1 | — | — | — | 41 | — | |
| PixelReasonerParam Size=7B2026.03 | 54.5 | 25.3 | — | 66.7 | — | 52.7 | — | — | — | 88.1 | — | — | — | 42.5 | — | |
| Qwen3-VL-2BParameter Scale=2B2026.05 | 53.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-7B + vanilla SFTParam Size=7B2026.03 | 53.1 | 26.7 | — | 60.8 | — | 45.3 | — | — | — | 88.8 | — | — | — | 33.6 | — | |
| InternVL3.5-2BParameter Scale=2B2026.05 | 51.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4V(ision)Access=API-based2026.02 | 51.14 | 22.67 | 78.52 | 60.83 | 59.68 | 70 | 79.49 | 26.15 | 28.78 | 72.73 | 54.92 | 33.72 | 55.64 | 38.81 | 34.09 | |
| GPT-4oParam Size=-2026.03 | 51.1 | 30 | — | 51.7 | — | 58 | — | — | — | 76.9 | — | — | — | 38.8 | — | |
| PerceptionLM-3BParameter Scale=3B2026.05 | 49.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Zamba2-VL-7BParameter Scale=7B2026.05 | 49.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LLaVA-v1.6-34BAccess=Open-source, Size=34B2026.02 | 46.8 | 26 | 48.89 | 66.67 | 67.74 | 54.67 | 43.59 | 20.77 | 23.74 | 74.83 | 59.02 | 30.81 | 62.41 | 31.34 | 44.7 | |
| PerceptionLM-1BParameter Scale=1B2026.05 | 45.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini ProAccess=API-based2026.02 | 45.16 | 23.33 | 52.59 | 52.5 | 40.32 | 57.33 | 50.43 | 24.62 | 26.62 | 74.83 | 53.28 | 42.44 | 44.36 | 38.81 | 50.76 | |
| Claude 3 OPUSAccess=API-based2026.02 | 44.05 | 24.67 | 72.59 | 50.83 | 47.58 | 32.67 | 65.81 | 21.54 | 25.18 | 58.04 | 51.64 | 36.63 | 56.39 | 26.87 | 46.21 | |
| InternVL3.5-1BParameter Scale=1B2026.05 | 43.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Zamba2-VL-1.2BParameter Scale=1.2B2026.05 | 43.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Vision-R1Param Size=7B2026.03 | 42.8 | 26.7 | — | 51.7 | — | 27.3 | — | — | — | 66.4 | — | — | — | 44.8 | — | |
| LLaVA-v1.5-13BAccess=Open-source, Size=13B2026.02 | 42.66 | 26 | 47.41 | 45 | 53.23 | 58 | 47.01 | 26.15 | 32.37 | 67.83 | 52.46 | 29.07 | 44.36 | 36.57 | 31.82 | |
| Zamba2-VL-2.7BParameter Scale=2.7B2026.05 | 42.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InstructBLIP-13BAccess=Open-source, Size=13B2026.02 | 42.24 | 24.67 | 49.63 | 30.83 | 51.61 | 52.67 | 51.28 | 29.23 | 32.37 | 65.73 | 55.74 | 29.65 | 57.14 | 38.81 | 21.97 | |
| LLaVA-v1.5-13B-xtunerAccess=Open-source, Size=13B, Training=xtuner2026.02 | 42 | 24.67 | 47.41 | 48.33 | 54.03 | 52 | 47.01 | 30 | 30.94 | 69.93 | 45.08 | 29.65 | 44.36 | 38.81 | 25.76 | |
| CogVLMAccess=Open-source2026.02 | 41.54 | 26.67 | 47.41 | 38.33 | 52.42 | 52.67 | 47.86 | 23.08 | 33.09 | 63.64 | 52.46 | 29.65 | 54.14 | 29.85 | 30.3 | |
| Qwen-VL-MaxAccess=API-based2026.02 | 40.28 | 21.33 | 51.11 | 56.67 | 58.06 | 4.67 | 38.46 | 28.46 | 23.02 | 69.93 | 48.36 | 31.4 | 51.88 | 36.57 | 43.94 | |
| InstructBLIP-7BAccess=Open-source, Size=7B2026.02 | 39.72 | 20 | 47.41 | 32.5 | 51.61 | 52.67 | 47.01 | 23.85 | 30.94 | 56.64 | 48.36 | 30.81 | 55.64 | 33.58 | 25 | |
| LLaVA-v1.5-7B-xtunerAccess=Open-source, Size=7B, Training=xtuner2026.02 | 39.36 | 18.67 | 47.41 | 45.83 | 51.61 | 52.67 | 47.01 | 20 | 28.78 | 68.53 | 36.89 | 29.07 | 38.35 | 29.85 | 36.36 | |
| OpenFlamingo-v2Access=Open-source2026.02 | 39.18 | 18.67 | 62.22 | 30 | 54.03 | 47.33 | 52.99 | 24.62 | 30.22 | 43.36 | 56.56 | 25.58 | 44.36 | 36.57 | 21.97 | |
| Random Choice2026.02 | 38.09 | 25 | 50 | 25 | 50 | 50 | 50 | 25 | 25 | 50 | 50 | 25 | 50 | 33.33 | 25 | |
| LLaVA-internLM2-7BAccess=Open-source, Size=7B2026.02 | 37.71 | 22 | 48.89 | 55 | 57.26 | 28.67 | 29.06 | 23.85 | 27.34 | 76.22 | 50 | 27.91 | 44.36 | 32.09 | 5.3 | |
| LLaVA-v1.5-7BAccess=Open-source, Size=7B2026.02 | 37.13 | 20 | 47.41 | 40 | 52.42 | 11.33 | 47.01 | 20 | 23.02 | 61.54 | 56.56 | 25.58 | 51.88 | 39.55 | 23.48 | |
| MiniGPT-4-v2Access=Open-source2026.02 | 34.23 | 20.67 | 44.44 | 13.33 | 50.81 | 34.67 | 43.59 | 20.77 | 28.78 | 44.76 | 47.54 | 26.16 | 48.87 | 30.6 | 24.24 | |
| PerceptionLM-8BParameter Scale=8B2026.05 | 27 | — | — | — | — | — | — | — | — | — | — | — | — | — | — |