Share your thoughts, 1 month free Claude Pro on usSee more

Spatial Reasoning on CrossPoint-Bench

91.75Score

Human

Updated 1mo ago

Evaluation Results

Method	Links
Human 2025.12		91.75
CroPond-7B 2025.12		76.8
CroPond-3B 2025.12		71.6
Qwen3-VL-235B-A22B-Thinking 2025.12		52.7
Qwen3-VL-30B-A3B-Thinking 2025.12		39.7
Gemini-2.5-Pro 2025.12		37.1
Qwen2.5-VL-72B-Instruct 2025.12		36.5
Qwen3-VL-235B-A22B-Instruct 2025.12		35.2
o3 2025.12		32.6
Qwen2.5-VL-32B-Instruct 2025.12		29.1
Qwen3-VL-30B-A3B-Instruct 2025.12		28.1
Qwen2.5-VL-7B-Instruct 2025.12		26.8
RoboBrain2.0-7B 2025.12		26
Molmo-7B-D 2025.12		25.8
Qwen2.5-VL-3B-Instruct 2025.12		25.4
RoboBrain2.0-7B 2025.12		25.4
Claude-Sonnet-4 2025.12		23.1
RoboRefer-8B-SFT 2025.12		21
RoboPoint-v1-Vicuna-13B 2025.12		14.6