Multimodal Mathematical Reasoning on MathVista (accuracy, Average)
67.5AccuracyMMS-PRM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MMS-PRMSize=8B2026.06 | 67.5 | 73 | |
| InternVL2.5-MPO + SFTSize=8B2026.06 | 65.9 | 71 | |
| InternVL2.5-MPOSize=8B2026.06 | 65 | 69.4 | |
| R1-OnevisionSize=7B2026.06 | 64.1 | — | |
| IXC-2.5Size=7B2026.06 | 63.7 | — | |
| R1-VLSize=7B2026.06 | 63.5 | — | |
| LLaVA-OneVisionSize=7B2026.06 | 63.2 | 64.6 | |
| Ovis1.5-LLaMA3Size=8B2026.06 | 63 | — | |
| Vision-R1-LlamaVSize=11B2026.06 | 62.7 | — | |
| InternVL-2.5Size=8B2026.06 | 62.2 | 68 | |
| MiniCPM-V-2-6Size=8B2026.06 | 60.6 | 64.2 | |
| Insight-VSize=7B2026.06 | 59.9 | 65.7 | |
| Qwen2-VLSize=7B2026.06 | 58.2 | 65.1 | |
| LLaVA-CoTSize=11B2026.06 | 54.8 | — | |
| LlamaV-o1Size=11B2026.06 | 54.4 | — | |
| LLaVA-NeXT-LLaMA3 + SFTSize=8B2026.06 | 51.4 | 60.8 | |
| LLaVA-NeXT-LLaMA3Size=8B2026.06 | 45.9 | 52.1 | |
| LLaVA-1.5Size=13B2026.06 | 27.6 | — |