Mathematics Reasoning on MathVision mini
60.54AccuracyOurs
Evaluation results
| Method | Links | |
|---|---|---|
| OursModel=Kimi-VL A3B-Thinking2025.10 | 60.54 | |
| AGLAModel=Kimi-VL A3B-Thinking2025.10 | 58.88 | |
| VanillaModel=Kimi-VL A3B-Thinking2025.10 | 56.24 | |
| VCDModel=Kimi-VL A3B-Thinking2025.10 | 56.14 | |
| PVM-8B (SFT + GRPO)Backbone=8B, Training Strategy=SFT + GRPO, Model Architecture=Persistent Visual Memory2026.05 | 51.3 | |
| PVM-8B (SFT)Backbone=8B, Training Strategy=SFT, Model Architecture=Persistent Visual Memory2026.05 | 50.7 | |
| CGDModel=Kimi-VL A3B-Thinking2025.10 | 49.3 | |
| Euclid-8BBackbone=8B, Training Strategy=RL-tuned2026.05 | 49 | |
| MemVRTraining Strategy=Visual Injection2026.05 | 48.4 | |
| ICoTTraining Strategy=Visual Injection2026.05 | 48.1 | |
| Qwen3-VL-8B (SFT)Backbone=8B, Training Strategy=SFT2026.05 | 48 | |
| Qwen3-VL-8B (SFT + GRPO)Backbone=8B, Training Strategy=SFT + GRPO2026.05 | 48 | |
| PEARL-8BBackbone=8B, Training Strategy=RL-tuned2026.05 | 47.7 | |
| Qwen3-VL-8B (LoRA-SFT + GRPO)Backbone=8B, Training Strategy=LoRA-SFT + GRPO2026.05 | 46.7 | |
| OneThinker-8BBackbone=8B, Training Strategy=RL-tuned2026.05 | 45.4 | |
| Qwen3-VL-8B-InstructBackbone=8B2026.05 | 45.4 | |
| PVM-4B (SFT + GRPO)Backbone=4B, Training Strategy=SFT + GRPO, Model Architecture=Persistent Visual Memory2026.05 | 45.4 | |
| CoMemoTraining Strategy=Visual Injection2026.05 | 43.4 | |
| Qwen3-VL-8B (LoRA-SFT)Backbone=8B, Training Strategy=LoRA-SFT2026.05 | 42.8 | |
| Qwen3-VL-4B (SFT + GRPO)Backbone=4B, Training Strategy=SFT + GRPO2026.05 | 42.8 | |
| Qwen3-VL-4B (LoRA-SFT + GRPO)Backbone=4B, Training Strategy=LoRA-SFT + GRPO2026.05 | 42.4 | |
| PVM-4B (SFT)Backbone=4B, Training Strategy=SFT, Model Architecture=Persistent Visual Memory2026.05 | 41.5 | |
| OursModel=R1-Onevision 7B2025.10 | 39.12 | |
| Qwen3-VL-4B (SFT)Backbone=4B, Training Strategy=SFT2026.05 | 37.5 | |
| Qwen3-VL-4B (LoRA-SFT)Backbone=4B, Training Strategy=LoRA-SFT2026.05 | 37.2 | |
| AGLAModel=R1-Onevision 7B2025.10 | 37.03 | |
| Qwen3-VL-4B-InstructBackbone=4B2026.05 | 35.9 | |
| VanillaModel=R1-Onevision 7B2025.10 | 33.54 | |
| VCDModel=R1-Onevision 7B2025.10 | 33.44 | |
| CGDModel=R1-Onevision 7B2025.10 | 28.12 | |
| OursModel=Ocean-R1 7B Instruct2025.10 | 26.01 | |
| AGLAModel=Ocean-R1 7B Instruct2025.10 | 23.69 | |
| VanillaModel=Ocean-R1 7B Instruct2025.10 | 20.05 | |
| VCDModel=Ocean-R1 7B Instruct2025.10 | 19.92 | |
| CGDModel=Ocean-R1 7B Instruct2025.10 | 10.4 |