CORE: Coarse-to-Fine Search with Context Management for Deep Research
Hoang Tran Vuong, Ngoc Quan Pham, Hung Bui, Fatih Porikli and Dat Quoc Nguyen (Qualcomm AI Research) present CORE, a training-free deep research harness that searches over long passages but returns only short snippets to the agent.
Ask this paper
Problem. Retrieval matches best over long passages, but every observation returned to the agent stays in its context for the rest of the trajectory.
Coarse-to-fine search. A dense retriever fetches passages, then token-level similarities from a late-interaction reranker select the most relevant snippet in each one. A retrieval model locates the evidence instead of the agent.
Context management. The agent can expand the context around sources it has already retrieved, and repeated snippets are suppressed.
Results. In closed-search comparisons with the Tongyi DeepResearch and OpenResearcher harnesses, CORE gets the best answer accuracy and evidence recall across three agent models and three benchmarks (BrowseComp-Plus, BrowseComp-Wikipedia, BrowseComp-Plus-CM).
Scale. The evaluation corpora contain up to 553M documents.
Abstract
Deep research agents answer complex questions through repeated search. They rely on a harness that runs their tool calls and returns evidence to their context. This creates a trade-off: Search works best over long passages, which give enough context for reliable semantic matching. But the returned evidence should be short, because every observation stays in the agent's context for the rest of the trajectory. We introduce CORE, a training-free harness that combines coarse-to-fine search with context management. CORE retrieves passages with a dense retriever. It then reuses the token-level similarities from a late-interaction reranker to return only the most relevant snippet from each passage. Context management lets the agent expand the context around sources it has already retrieved and suppresses repeated snippets. In controlled closed-search comparisons with the Tongyi DeepResearch and OpenResearcher harnesses, CORE achieves state-of-the-art answer accuracy and evidence recall across three agent models and three benchmarks (BrowseComp-Plus, BrowseComp-Wikipedia and BrowseComp-Plus-CM), on corpora containing up to 553M documents.