Alibaba Cloud Tongyi has open-sourced a new visual model Qwen2.5-VL, launching three sizes: 3B, 7B, and 72B. The flagship model Qwen2.5-VL-72B won the championship in visual understanding across 13 authoritative evaluations, surpassing GPT-4o and Claude 3.5. According to Alibaba Cloud's official introduction, the new Qwen2.5-VL can analyze image content more accurately and support breakthrough video understanding of over 1 hour. This model can search for specific events in videos, and provide insights on...