Search-o1

RUC-NLPIR's (Zhicheng Dou) agentic search-enhanced large reasoning models (EMNLP 2025; ~450 citations, 1.2K★): an agentic RAG loop plus a Reason-in-Documents module lets o1-style reasoners retrieve mid-chain without breaking coherence — the reference point for the 2025 search-agent wave.