阿里千问视觉模型登顶全球空间推理榜，超越Gemini3和GPT5.1

Core Insights - Alibaba's Qwen models have achieved top rankings in the latest SpatialBench benchmark for spatial reasoning, surpassing leading international models like Gemini 3 and GPT-5.1 [1] Model Performance - Qwen3-VL scored 13.5 points and Qwen2.5-VL scored 12.9 points, significantly outperforming Gemini 3.0 Pro Preview (9.6), GPT-5.1 (7.5), and Claude Sonnet 4.5 [1] - Despite the strong performance of AI models, there remains a notable gap compared to human capabilities, with the human benchmark around 80 points for complex spatial reasoning tasks [1] Model Capabilities - Qwen3-VL excels in recognition, multi-target grounding, and understanding spatial relations, indicating advanced capabilities in visual understanding [4]