Geometry problem solving on MathVista GPS
95.7AccuracyDreamPRM (o4-mini)
Evaluation Results
| Method | Links | |
|---|---|---|
| DreamPRM (o4-mini)Model Access Type=Close-Source2025.05 | 95.7 | |
| Kimi-k1.6-preview-20250308Model Access Type=Close-Source2025.05 | 91.8 | |
| Step R1-V-MiniModel Access Type=Close-Source2025.05 | 89.9 | |
| Doubao-pro-1.5Model Access Type=Close-Source2025.05 | 88.9 | |
| AutoGPSModel Access Type=Open-Source, Base Model=InternVL32025.05 | 86.2 | |
| Ovis2_34BModel Access Type=Open-Source2025.05 | 84.6 | |
| Vision-R1-7BModel Access Type=Open-Source2025.05 | 82.7 | |
| MathCoder-VL-8B#Params=8B2025.05 | 73.6 | |
| InternVL2-76B#Params=76B2025.05 | 67.8 | |
| Multimath-7B#Params=7B2025.05 | 66.8 | |
| MathCoder-VL-2B#Params=2B2025.05 | 66.4 | |
| GPT-4o#Params=-2025.05 | 64.7 | |
| Claude3.5-Sonnet#Params=-2025.05 | 64.4 | |
| MathGLM-Vision-9B#Params=9B2025.05 | 64.4 | |
| MAVIS-7B#Params=7B2025.05 | 64.1 | |
| IXC-2.5-Reward#Params=7B2025.05 | 63.5 | |
| IXC-2-VL#Params=7B2025.05 | 63 | |
| InternVL2-8B#Params=8B2025.05 | 62 | |
| Gemini-1.5-Pro#Params=-2025.05 | 58.9 | |
| GPT-4-turbo#Params=-2025.05 | 58.3 | |
| Math-LLaVA-13B#Params=13B2025.05 | 57.7 | |
| InternVL2-26B#Params=26B2025.05 | 54.3 | |
| Claude3-Opus#Params=-2025.05 | 52.9 | |
| GPT-4V#Params=-2025.05 | 50.5 | |
| G-LLaVA-7B#Params=7B2025.05 | 48.7 | |
| Human2025.05 | 48.4 | |
| Human2025.05 | 48.4 | |
| Math-PUMA-Qwen2#Params=8B2025.05 | 48.1 | |
| Qwen-VL-Max#Params=-2025.05 | 46.1 | |
| Qwen2-VL#Params=8B2025.05 | 40.9 | |
| Math-PUMA-DS#Params=7B2025.05 | 39.9 | |
| InternVL-Chat-2B-V1-5#Params=2B2025.05 | 37.5 | |
| Qwen-VL-Plus#Params=-2025.05 | 35.5 | |
| InfiMM-Math#Params=7B2025.05 | 34.5 | |
| Deepseek-VL#Params=8B2025.05 | 28.4 | |
| LLaVA-1.5-13B#Params=13B2025.05 | 22.7 | |
| Random Chance2025.05 | 21.6 | |
| SPHINX-V2-13B#Params=13B2025.05 | 16.4 |