{"id":"https://openalex.org/W7161832508","doi":"https://doi.org/10.48550/arxiv.2605.18796","title":"UCCI: Calibrated Uncertainty for Cost-Optimal LLM Cascade Routing","display_name":"UCCI: Calibrated Uncertainty for Cost-Optimal LLM Cascade Routing","publication_year":2026,"publication_date":"2026-05-11","ids":{"openalex":"https://openalex.org/W7161832508","doi":"https://doi.org/10.48550/arxiv.2605.18796"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.18796","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.18796","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.18796","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136577318","display_name":"Varun Kotte","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Kotte, Varun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5136577318"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12326","display_name":"Network Packet Processing and Optimization","score":0.5853000283241272,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T12326","display_name":"Network Packet Processing and Optimization","score":0.5853000283241272,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10714","display_name":"Software-Defined Networks and 5G","score":0.15610000491142273,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T12292","display_name":"Graph Theory and Algorithms","score":0.061400000005960464,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/calibration","display_name":"Calibration","score":0.5893999934196472},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5710999965667725},{"id":"https://openalex.org/keywords/routing","display_name":"Routing (electronic design automation)","score":0.5512999892234802},{"id":"https://openalex.org/keywords/margin","display_name":"Margin (machine learning)","score":0.5493999719619751},{"id":"https://openalex.org/keywords/cascade","display_name":"Cascade","score":0.48660001158714294},{"id":"https://openalex.org/keywords/router","display_name":"Router","score":0.4812999963760376},{"id":"https://openalex.org/keywords/workload","display_name":"Workload","score":0.45879998803138733}],"concepts":[{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.5893999934196472},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5848000049591064},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5710999965667725},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.5512999892234802},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.5493999719619751},{"id":"https://openalex.org/C34146451","wikidata":"https://www.wikidata.org/wiki/Q5048094","display_name":"Cascade","level":2,"score":0.48660001158714294},{"id":"https://openalex.org/C2775896111","wikidata":"https://www.wikidata.org/wiki/Q642560","display_name":"Router","level":2,"score":0.4812999963760376},{"id":"https://openalex.org/C2778476105","wikidata":"https://www.wikidata.org/wiki/Q628539","display_name":"Workload","level":2,"score":0.45879998803138733},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.3659000098705292},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.3395000100135803},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3190999925136566},{"id":"https://openalex.org/C17418463","wikidata":"https://www.wikidata.org/wiki/Q3455874","display_name":"Isotonic regression","level":3,"score":0.2922999858856201},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.27160000801086426},{"id":"https://openalex.org/C44249647","wikidata":"https://www.wikidata.org/wiki/Q208498","display_name":"Confidence interval","level":2,"score":0.27000001072883606},{"id":"https://openalex.org/C50817715","wikidata":"https://www.wikidata.org/wiki/Q79895177","display_name":"Regret","level":2,"score":0.2671999931335449},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.2533000111579895},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.25200000405311584},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.25029999017715454},{"id":"https://openalex.org/C184896649","wikidata":"https://www.wikidata.org/wiki/Q290066","display_name":"Routing table","level":4,"score":0.2500999867916107}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.18796","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.18796","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.18796","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.18796","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"LLM":[0],"cascades":[1],"and":[2,17,33,57,78,103,142,157],"model":[3,16,155],"routing":[4,152,163],"promise":[5],"lower":[6],"inference":[7,112],"cost":[8,64,113],"by":[9,62,101,114],"sending":[10],"easy":[11],"queries":[12,99],"to":[13,21,49,129],"a":[14,22,41,50,91,143],"small":[15],"escalating":[18],"hard":[19],"ones":[20],"large":[23],"model,":[24],"but":[25],"most":[26],"deployed":[27],"routers":[28],"use":[29,150],"uncalibrated":[30],"confidence":[31],"scores":[32],"require":[34],"per-workload":[35],"threshold":[36,61,70],"tuning.":[37],"We":[38],"present":[39],"UCCI,":[40],"calibration-first":[42],"router":[43],"that":[44],"maps":[45],"token-level":[46],"margin":[47],"uncertainty":[48],"per-query":[51],"error":[52,88],"probability":[53],"via":[54],"isotonic":[55,79],"regression":[56],"selects":[58],"the":[59,73,132],"escalation":[60],"constrained":[63],"minimization.":[65],"Under":[66],"three":[67],"explicit":[68],"assumptions,":[69],"policies":[71],"on":[72,107,153],"calibrated":[74],"score":[75],"are":[76],"cost-optimal,":[77],"calibration":[80,87],"achieves":[81],"O(n^{-1/3})":[82],"sample":[83],"complexity":[84],"for":[85],"expected":[86],"(ECE).":[89],"On":[90],"production":[92],"named":[93],"entity":[94],"recognition":[95],"workload":[96],"of":[97],"75,000":[98],"served":[100],"4B":[102],"12B":[104],"instruction-tuned":[105],"LLMs":[106],"H100":[108,159],"GPUs,":[109],"UCCI":[110,136],"cuts":[111],"31%":[115],"(95%":[116],"CI:":[117],"[27%,":[118],"35%])":[119],"at":[120],"micro-F1":[121],"=":[122],"0.91":[123],"while":[124],"reducing":[125],"ECE":[126],"from":[127,164],"0.12":[128],"0.03.":[130],"At":[131],"same":[133],"operating":[134],"point,":[135],"beats":[137],"entropy":[138],"thresholding,":[139],"split-conformal":[140],"routing,":[141],"FrugalGPT-style":[144],"learned":[145],"threshold.":[146],"All":[147],"cascade":[148],"results":[149],"end-to-end":[151],"actual":[154],"outputs":[156],"measured":[158],"latency,":[160],"not":[161],"simulated":[162],"global":[165],"accuracies":[166],"or":[167],"nominal":[168],"API":[169],"prices.":[170]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-21T00:00:00"}