Spatial Reasoning on CrossPoint-Bench
91.75ScoreHuman
Evaluation Results
| Method | Links | |
|---|---|---|
| Human2025.12 | 91.75 | |
| CroPond-7BParameters=7B2025.12 | 76.8 | |
| CroPond-3BParameters=3B2025.12 | 71.6 | |
| Qwen3-VL-235B-A22B-ThinkingParameters=235B, MoE=A22B, Mode=Thinking2025.12 | 52.7 | |
| Qwen3-VL-30B-A3B-ThinkingParameters=30B, MoE=A3B, Mode=Thinking2025.12 | 39.7 | |
| Gemini-2.5-Pro2025.12 | 37.1 | |
| Qwen2.5-VL-72B-InstructParameters=72B, Mode=Instruct2025.12 | 36.5 | |
| Qwen3-VL-235B-A22B-InstructParameters=235B, MoE=A22B, Mode=Instruct2025.12 | 35.2 | |
| o32025.12 | 32.6 | |
| Qwen2.5-VL-32B-InstructParameters=32B, Mode=Instruct2025.12 | 29.1 | |
| Qwen3-VL-30B-A3B-InstructParameters=30B, MoE=A3B, Mode=Instruct2025.12 | 28.1 | |
| Qwen2.5-VL-7B-InstructParameters=7B, Mode=Instruct2025.12 | 26.8 | |
| RoboBrain2.0-7BParameters=7B, reasoning_mode=thinking2025.12 | 26 | |
| Molmo-7B-DParameters=7B2025.12 | 25.8 | |
| Qwen2.5-VL-3B-InstructParameters=3B, Mode=Instruct2025.12 | 25.4 | |
| RoboBrain2.0-7BParameters=7B, reasoning_mode=no-thinking2025.12 | 25.4 | |
| Claude-Sonnet-42025.12 | 23.1 | |
| RoboRefer-8B-SFTParameters=8B, Mode=SFT2025.12 | 21 | |
| RoboPoint-v1-Vicuna-13BBackbone=Vicuna-13B2025.12 | 14.6 |