The field of video question answering and generation is moving towards more comprehensive and nuanced understanding of video content. Researchers are exploring new approaches to enhance the reasoning capability of video question answering models, such as generating question-answer pairs from descriptive information extracted directly from videos and aligning task-specific question embeddings with corresponding visual features. Additionally, there is a trend towards developing more efficient and lightweight video generation models that can achieve state-of-the-art performance with reduced parameters. The use of scene graphs, graph neural networks, and cross-modality proxy queries are also being investigated to improve the accuracy and interpretability of video question answering and object segmentation models. Noteworthy papers include: FIQ, which introduces a framework to enhance the reasoning capability of VQA models by improving their foundational comprehension of video content. HunyuanVideo 1.5, which presents a lightweight yet powerful open-source video generation model that achieves state-of-the-art visual quality and motion coherence. SFA, which proposes a training-free framework for Video TextVQA that effectively guides the Video-LLM's attention toward essential cues. GHR-VQA, which incorporates scene graphs to capture intricate human-object interactions within video sequences. ProxyFormer, which introduces a set of proxy queries to integrate visual and text semantics and facilitate the flow of semantics between them.