{"id":"https://openalex.org/W7166863299","doi":"https://doi.org/10.18653/v1/2026.findings-acl.994","title":"X-Router: Decoupling Knowledge and Reasoning for Cost-Effective LLM Inference","display_name":"X-Router: Decoupling Knowledge and Reasoning for Cost-Effective LLM Inference","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166863299","doi":"https://doi.org/10.18653/v1/2026.findings-acl.994"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.994","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.994","pdf_url":"https://aclanthology.org/2026.findings-acl.994.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.994.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139766814","display_name":"Zixuan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zixuan Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139732992","display_name":"Yinze Ding","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yinze Ding","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139850535","display_name":"Zihan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zihan Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139832632","display_name":"Jinyu Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jinyu Guo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139844648","display_name":"Zhenhong Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhenhong Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139754633","display_name":"Junhao Dong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Junhao Dong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5020410767","display_name":"Chaomeng Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chaomeng Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.85219172,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"19856","last_page":"19874"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10215","display_name":"Semantic Web and Ontologies","score":0.12680000066757202,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10215","display_name":"Semantic Web and Ontologies","score":0.12680000066757202,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.05490000173449516,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.04349999874830246,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.4672999978065491},{"id":"https://openalex.org/keywords/decoupling","display_name":"Decoupling (probability)","score":0.3382999897003174},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.335999995470047},{"id":"https://openalex.org/keywords/non-monotonic-logic","display_name":"Non-monotonic logic","score":0.3330000042915344},{"id":"https://openalex.org/keywords/knowledge-representation-and-reasoning","display_name":"Knowledge representation and reasoning","score":0.3149000108242035}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5945000052452087},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4887000024318695},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4672999978065491},{"id":"https://openalex.org/C205606062","wikidata":"https://www.wikidata.org/wiki/Q5249645","display_name":"Decoupling (probability)","level":2,"score":0.3382999897003174},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.335999995470047},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.3330000042915344},{"id":"https://openalex.org/C161301231","wikidata":"https://www.wikidata.org/wiki/Q3478658","display_name":"Knowledge representation and reasoning","level":2,"score":0.3149000108242035},{"id":"https://openalex.org/C115925183","wikidata":"https://www.wikidata.org/wiki/Q1412694","display_name":"Knowledge-based systems","level":2,"score":0.2858999967575073},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.28349998593330383},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2531999945640564}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.994","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.994","pdf_url":"https://aclanthology.org/2026.findings-acl.994.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.994","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.994","pdf_url":"https://aclanthology.org/2026.findings-acl.994.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166863299.pdf","grobid_xml":"https://content.openalex.org/works/W7166863299.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"are":[4,36],"often":[5],"augmented":[6],"with":[7],"Retrieval-Augmented":[8],"Generation":[9],"(RAG)":[10],"and":[11,33,66,81,96,125],"Chain-of-Thought":[12],"(CoT)":[13],"prompting,":[14],"yet":[15],"static":[16],"\"always-on\"":[17],"use":[18],"is":[19],"computationally":[20],"wasteful.Existing":[21],"adaptive":[22],"methods":[23],"typically":[24],"optimize":[25],"a":[26,42,54,83],"single":[27],"axis,":[28],"overlooking":[29],"that":[30,46,74],"evidence":[31],"need":[32],"reasoning":[34,51],"depth":[35],"only":[37],"partially":[38],"correlated.We":[39],"present":[40],"X-ROUTER,":[41],"dual-axis":[43],"routing":[44],"framework":[45],"separates":[47],"retrieval":[48],"necessity":[49,52],"from":[50],"under":[53],"user-defined":[55],"cost-quality":[56],"trade-off.Offline,":[57],"X-ROUTER":[58,117],"profiles":[59],"four":[60],"pipelines":[61],"(DIRECT,":[62],"RAG,":[63],"COT,":[64],"RAG+COT)":[65],"derives":[67],"supervision":[68],"by":[69,121,127],"selecting":[70],"the":[71],"utilitymaximizing":[72],"strategy":[73],"trades":[75],"answer":[76,133],"quality":[77,134],"against":[78],"token":[79,119],"usage":[80,120],"latency.Online,":[82],"compact":[84],"dual-head":[85],"router,":[86],"conditioned":[87],"on":[88],"cost":[89],"weights,":[90],"uses":[91],"lightweight":[92],"probesretrieval-score":[93],"dispersion":[94],"(NQC)":[95],"singlepass":[97],"draft":[98],"negative":[99],"log-likelihood":[100],"(NLL)-to":[101],"decide":[102],"whether":[103],"to":[104,123,129],"invoke":[105],"RAG":[106],"and/or":[107],"CoT":[108],"without":[109],"sampling":[110],"or":[111],"model":[112],"internals.Across":[113],"six":[114],"QA":[115],"benchmarks,":[116],"reduces":[118],"up":[122,128],"86%":[124],"latency":[126],"84%":[130],"while":[131],"improving":[132],"over":[135],"strong":[136],"baselines.":[137]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}