@inproceedings{wang-etal-2026-x,
    title = "{X}-Router: Decoupling Knowledge and Reasoning for Cost-Effective {LLM} Inference",
    author = "Wang, Zixuan  and
      Ding, Yinze  and
      Wang, Zihan  and
      Guo, Jinyu  and
      Zhou, Zhenhong  and
      Dong, Junhao  and
      Chen, Chaomeng",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.994/",
    pages = "19856--19874",
    ISBN = "979-8-89176-395-1",
    abstract = "Large Language Models (LLMs) are often augmented with Retrieval-Augmented Generation (RAG) and Chain-of-Thought (CoT) prompting, yet static ``always-on'' use is computationally wasteful. Existing adaptive methods typically optimize a single axis, overlooking that evidence need and reasoning depth are only partially correlated. We present , a dual-axis routing framework that separates retrieval necessity from reasoning necessity under a user-defined cost{--}quality trade-off. Offline, profiles four pipelines (Direct, RAG, CoT, RAG+CoT) and derives supervision by selecting the utility-maximizing strategy that trades answer quality against token usage and latency. Online, a compact dual-head router, conditioned on cost weights, uses lightweight probes{---}retrieval-score dispersion (NQC) and single-pass draft negative log-likelihood (NLL){---}to decide whether to invoke RAG and/or CoT without sampling or model internals. Across six QA benchmarks, reduces token usage by up to 86{\%} and latency by up to 84{\%} while improving answer quality over strong baselines."
}