Multimodal Reasoning on MathVista (Strict Accuracy, Standard Accuracy, False Positive)
73.6Strict AccuracyAutoRubric
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AutoRubric2025.10 | 73.6 | 75.8 | -2.2 | |
| NoisyRollout2025.10 | 71.5 | 73 | -1.5 | |
| Perception-R12025.10 | 71.4 | 72 | -0.6 | |
| VL-Rethinker2025.10 | 70.8 | 73.9 | -3.1 | |
| VL-Reasoner2025.10 | 70.6 | 74.8 | -4.2 | |
| Vision-G12025.10 | 70 | 76.4 | -6.4 | |
| MM-Eureka2025.10 | 69.8 | 71.8 | -2 | |
| ThinkLite-VL2025.10 | 68.7 | 73.3 | -4.6 | |
| VLAA-Thinker2025.10 | 67.7 | 70.1 | -2.4 | |
| Qwen2.5-VL-7B2025.10 | 66 | 68.6 | -2.6 | |
| R1-VL2025.10 | 53 | 54.9 | -1.9 |