Aggregate Multimodal Performance on Average
80ScoreQwen2.5-VL-72B + CARES
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-72B + CARESCost=-70%2025.10 | 80 | |
| Qwen2.5-VL-72B2025.10 | 79 | |
| InternVL3-8B2025.10 | 77 | |
| InternVL3-8B + CARESCost=-64%2025.10 | 77 | |
| InternVL3-8B + CARES-ARCost=-76%2025.10 | 76 | |
| GPT-4o2025.10 | 69 | |
| GPT-4o + CARESCost=-55%2025.10 | 68 | |
| Granite-Vision -2B + CARESCost=-63%2025.10 | 60 | |
| Granite-Vision -2B2025.10 | 59 | |
| Granite-Vision -2B + CARES-ARCost=-67%2025.10 | 58 |