{"id":"https://openalex.org/W7171516088","doi":"https://doi.org/10.48550/arxiv.2607.23933","title":"SpecBox: Speculative Sandbox Scheduling for Efficient LLM Agent Serving","display_name":"SpecBox: Speculative Sandbox Scheduling for Efficient LLM Agent Serving","publication_year":2026,"publication_date":"2026-07-27","ids":{"openalex":"https://openalex.org/W7171516088","doi":"https://doi.org/10.48550/arxiv.2607.23933"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2607.23933","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.23933","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":null,"license_id":null,"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.2607.23933","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100683233","display_name":"Yihui Zhang","orcid":"https://orcid.org/0000-0003-0885-2067"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhang, Yihui","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066123431","display_name":"Tianyu Wo","orcid":"https://orcid.org/0000-0002-5331-3364"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wo, Tianyu","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5115028374","display_name":"Jinghao Wang","orcid":"https://orcid.org/0009-0009-5919-5328"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wang, Jinghao","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023692113","display_name":"Xiaoyang Sun","orcid":"https://orcid.org/0000-0001-9396-739X"},"institutions":[{"id":"https://openalex.org/I130828816","display_name":"University of Leeds","ror":"https://ror.org/024mrxd33","country_code":"GB","type":"education","lineage":["https://openalex.org/I130828816"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Sun, Xiaoyang","raw_affiliation_strings":["University of Leeds"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Leeds","institution_ids":["https://openalex.org/I130828816"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5031176052","display_name":"Menghao Zhang","orcid":"https://orcid.org/0000-0001-5274-5512"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhang, Menghao","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113730514","display_name":"Cangzhou Yuan","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuan, Cangzhou","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143819360","display_name":"Li Li","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Li, Li","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143860648","display_name":"Chunming Hu","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hu, Chunming","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143805281","display_name":"Albert Y. Zomaya","orcid":null},"institutions":[{"id":"https://openalex.org/I129604602","display_name":"The University of Sydney","ror":"https://ror.org/0384j8v12","country_code":"AU","type":"education","lineage":["https://openalex.org/I129604602"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Zomaya, Albert Y.","raw_affiliation_strings":["The University of Sydney"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Sydney","institution_ids":["https://openalex.org/I129604602"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5143822034","display_name":"Renyu Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yang, Renyu","raw_affiliation_strings":["Beihang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":3,"corresponding_author_ids":[],"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/T12203","display_name":"Mobile Agent-Based Network Management","score":0.2969000041484833,"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"}},"topics":[{"id":"https://openalex.org/T12203","display_name":"Mobile Agent-Based Network Management","score":0.2969000041484833,"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/T12127","display_name":"Software System Performance and Reliability","score":0.1712000072002411,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.051899999380111694,"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/sandbox","display_name":"Sandbox (software development)","score":0.9860000014305115},{"id":"https://openalex.org/keywords/handshake","display_name":"Handshake","score":0.5378000140190125},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.44920000433921814},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.40470001101493835},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.39640000462532043},{"id":"https://openalex.org/keywords/cryptocurrency","display_name":"Cryptocurrency","score":0.3560999929904938},{"id":"https://openalex.org/keywords/planetlab","display_name":"PlanetLab","score":0.33570000529289246},{"id":"https://openalex.org/keywords/bootstrapping","display_name":"Bootstrapping (finance)","score":0.313400000333786}],"concepts":[{"id":"https://openalex.org/C167981075","wikidata":"https://www.wikidata.org/wiki/Q2667186","display_name":"Sandbox (software development)","level":2,"score":0.9860000014305115},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8198999762535095},{"id":"https://openalex.org/C2778000800","wikidata":"https://www.wikidata.org/wiki/Q830043","display_name":"Handshake","level":3,"score":0.5378000140190125},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.44920000433921814},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.4413999915122986},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.40470001101493835},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.39640000462532043},{"id":"https://openalex.org/C180706569","wikidata":"https://www.wikidata.org/wiki/Q13479982","display_name":"Cryptocurrency","level":2,"score":0.3560999929904938},{"id":"https://openalex.org/C2780300890","wikidata":"https://www.wikidata.org/wiki/Q851973","display_name":"PlanetLab","level":3,"score":0.33570000529289246},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.3273000121116638},{"id":"https://openalex.org/C207609745","wikidata":"https://www.wikidata.org/wiki/Q4944086","display_name":"Bootstrapping (finance)","level":2,"score":0.313400000333786},{"id":"https://openalex.org/C51929080","wikidata":"https://www.wikidata.org/wiki/Q2425187","display_name":"Codebase","level":3,"score":0.30709999799728394},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3028999865055084},{"id":"https://openalex.org/C194541083","wikidata":"https://www.wikidata.org/wiki/Q457174","display_name":"Workaround","level":2,"score":0.30169999599456787},{"id":"https://openalex.org/C15569618","wikidata":"https://www.wikidata.org/wiki/Q3561421","display_name":"Liveness","level":2,"score":0.29980000853538513},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.29829999804496765},{"id":"https://openalex.org/C147297375","wikidata":"https://www.wikidata.org/wiki/Q6674930","display_name":"Look-ahead","level":2,"score":0.2939000129699707},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.28459998965263367},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.28119999170303345},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.27090001106262207},{"id":"https://openalex.org/C2779010991","wikidata":"https://www.wikidata.org/wiki/Q2720909","display_name":"Artifact (error)","level":2,"score":0.2565999925136566}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2607.23933","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.23933","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2607.23933","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.23933","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":null,"license_id":null,"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":{"As":[0],"LLM":[1,74],"agents":[2],"increasingly":[3],"rely":[4],"on":[5,176],"the":[6,119,196],"Model":[7],"Context":[8],"Protocol":[9],"(MCP)":[10],"to":[11,89,130,170,192,195,208],"invoke":[12],"isolated":[13],"external":[14],"sandboxes,":[15],"disaggregated":[16],"sandbox":[17,32,69,92,106,127,134,156,198,211],"deployment":[18],"introduces":[19],"a":[20,64,126,148,159],"fundamental":[21],"tension":[22],"between":[23],"resource":[24],"utilization":[25],"and":[26,85,103,158],"interactive":[27],"tail":[28],"latency.":[29],"Persistent":[30],"long-lived":[31],"reservations":[33],"incur":[34],"excessive":[35],"memory":[36,203],"overhead":[37],"at":[38],"scale,":[39],"while":[40,200],"lazy":[41],"on-demand":[42,197],"instantiation":[43],"generates":[44],"severe":[45],"cold-start":[46],"penalties":[47],"that":[48,152,165,184],"degrade":[49],"response":[50],"performance":[51],"under":[52],"multi-tenant,":[53],"multi-turn":[54,178],"agent":[55,75,117,179],"workloads.":[56],"To":[57,111],"resolve":[58],"this":[59],"dilemma,":[60],"we":[61],"present":[62],"SpecBox,":[63],"runtime":[65],"built":[66],"around":[67],"speculative":[68,142],"preallocation":[70],"tailored":[71],"for":[72],"dynamic":[73],"execution":[76,98],"pipelines.":[77],"At":[78],"its":[79],"core,":[80],"SpecBox":[81,185],"implements":[82],"keyword":[83],"matching":[84],"streaming":[86],"semantic":[87,149],"embedding":[88],"enable":[90],"intent-driven":[91],"prewarming,":[93],"which":[94],"identifies":[95],"pending":[96],"tool":[97],"demands":[99],"mid-LLM":[100],"token":[101],"generation":[102],"fully":[104],"overlaps":[105],"bootstrapping":[107],"with":[108,144],"model":[109],"inference.":[110],"extend":[112],"prewarming":[113],"windows":[114],"across":[115],"sequential":[116],"steps,":[118],"framework":[120],"leverages":[121],"context-aware":[122],"stochastic":[123],"prefetching":[124],"atop":[125],"dependency":[128],"graph":[129],"probabilistically":[131],"forecast":[132],"future":[133],"switches":[135],"ahead":[136],"of":[137],"execution.":[138],"We":[139],"complement":[140],"these":[141],"mechanisms":[143],"two":[145],"orthogonal":[146],"optimizations:":[147],"result":[150],"cache":[151],"prunes":[153],"redundant":[154],"repeated":[155],"invocations,":[157],"dedicated":[160],"out-of-band":[161],"shared-memory":[162],"transport":[163],"plane":[164],"bypasses":[166],"conventional":[167],"network":[168],"serialization":[169],"deliver":[171],"zero-copy":[172],"artifact":[173],"transfers.":[174],"Evaluated":[175],"high-concurrency":[177],"traces,":[180],"our":[181],"prototype":[182],"demonstrates":[183],"cuts":[186],"P99":[187],"end-to-end":[188],"latency":[189],"by":[190,205],"up":[191],"$2.9\\times$":[193],"relative":[194],"baseline,":[199],"slashing":[201],"peak":[202],"consumption":[204],"$45.9\\%$":[206],"compared":[207],"permanently":[209],"reserved":[210],"deployments.":[212]},"counts_by_year":[],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2026-07-29T00:00:00"}