VideoResearchAgent: Grounded Task Synthesis and Sim-to-Real RL for Open-Web Video Research

Yuhang Zhou, Fei Li, Yuxi Wu +2

Existing deep research agents are designed primarily for text- and image-based web sources, while video reasoning systems typically assume that relevant videos are provided in adva…