Qwen2 VL

Alibaba open vision-language model with dynamic resolution and video understanding.