Low-level Visual Reasoning on BLINK
72.3AccuracyMasters
Evaluation Results
| Method | Links | |
|---|---|---|
| MastersBase Model=Qwen3-VL-8B2025.12 | 72.3 | |
| GPT-4.12025.12 | 68.5 | |
| GPT-4o2025.12 | 68 | |
| MastersBase Model=InternVL3-8B2025.12 | 68 | |
| MastersBase Model=InternVL3.5-8B2025.12 | 67.8 | |
| Qwen3-VL-32B2025.12 | 67.3 | |
| MastersBase Model=Qwen2.5-VL-7B2025.12 | 67.2 | |
| InternVL3-78B2025.12 | 66.3 | |
| GLM-4.5V2025.12 | 65.3 | |
| Qwen2.5-VL-72B2025.12 | 64.4 | |
| Gemini-2.0-Flash2025.12 | 64 | |
| InternVL3.5-38B2025.12 | 60.9 | |
| Claude-4-Sonnet2025.12 | 60.4 | |
| Claude-3.5-Sonnet2025.12 | 60.1 | |
| Gemini-1.5-Pro2025.12 | 59.1 | |
| Claude-3.7-Sonnet2025.12 | 56.6 | |
| LLaVA-OneVision-72B2025.12 | 55.4 | |
| Molmo-72B2025.12 | 49.7 | |
| NVLM-72B2025.12 | 48 |