DeepSeek V4-Flash Vision Video RAG lets AI truly understand a video and answer your questions: it tells you the answer, the exact minute and second where it occurs, and extracts a playable clip and keyframe for verification. Based on the DeepSeek vision model deepseek-v4-flash-vision-exp, this video understanding and question-answering (video RAG) agent skill first samples frames along the timeline and builds an index (once), then performs local coarse filtering → visual reranking → in-depth ans
liangdabiao/deepseek-v4-flash-vision-video-rag
Skill Claude CodeCodex
A video-understanding and question-answering tool based on a DeepSeek vision model. It can examine videos such as recordings, surveillance footage, games, films, or animations and refer to events with timestamps.