Visual Mathematical Reasoning on DynaMath worst
34.9ScoreOpenMMReasoner
Evaluation results
| Method | Links | |
|---|---|---|
| OpenMMReasonerSize=7B2026.03 | 34.9 | |
| GPT-4oSize=-2026.03 | 34.5 | |
| Insight-V++ (Qwen2.5-VL + Self-Evolving)Size=7B2026.03 | 33.6 | |
| Open-Vision-ReasonerSize=7B2026.03 | 33.5 | |
| GPT-4o miniSize=-2026.03 | 31.6 | |
| Insight-V (Qwen2.5-VL + Multi-Agent RL)Size=7B2026.03 | 31.4 | |
| MMR1Size=7B2026.03 | 27.9 | |
| InternVL3-8BSize=8B2026.03 | 23 | |
| MM-Eureka-QwenSize=7B2026.03 | 23 | |
| VLAA-Thinker-Qwen2.5Size=7B2026.03 | 22.4 | |
| Qwen2.5-VLSize=7B2026.03 | 21.8 | |
| LLaVA-OneVision-1.5Size=8B2026.03 | 19.8 | |
| VL-RethinkerSize=7B2026.03 | 17.8 | |
| OpenVLThinkerSize=7B2026.03 | 16.8 | |
| ThinkLite-VLSize=7B2026.03 | 16.5 | |
| LLaVA-OneVisionSize=7B2026.03 | 9 |