{"id":"https://openalex.org/W4415314084","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1285","title":"Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models","display_name":"Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4415314084","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1285"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1285","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1285","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1285","pdf_url":"https://aclanthology.org/2026.findings-acl.1285.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2026.findings-acl.1285.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5024315328","display_name":"Gaurav K. Srivastava","orcid":"https://orcid.org/0000-0002-3488-6220","observed_orcids":["https://orcid.org/0000-0002-3488-6220"]},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gaurav Srivastava","raw_affiliation_strings":["Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Aafiya Shamshad Hussain","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Aafiya Shamshad Hussain","raw_affiliation_strings":["Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5077499610","display_name":"Sriram Srinivasan","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Sriram Srinivasan","raw_affiliation_strings":["Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101476643","display_name":"Xuan Wang","orcid":"https://orcid.org/0009-0007-6541-9733","observed_orcids":["https://orcid.org/0009-0007-6541-9733"]},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xuan Wang","raw_affiliation_strings":["Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , Virginia Tech , Blacksburg , VA , USA,","institution_ids":["https://openalex.org/I859038795"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I859038795"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.01123285,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"25784","last_page":"25826"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.7832000255584717,"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/T10028","display_name":"Topic Modeling","score":0.7832000255584717,"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":0.08640000224113464,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.018799999728798866,"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/large-language-models","display_name":"large language models","score":0.9269999861717224},{"id":"https://openalex.org/keywords/overthinking","display_name":"overthinking","score":0.7480000257492065},{"id":"https://openalex.org/keywords/token-efficiency","display_name":"token efficiency","score":0.8629999756813049},{"id":"https://openalex.org/keywords/catastrophic-collapse","display_name":"catastrophic collapse","score":0.7149999737739563},{"id":"https://openalex.org/keywords/quantized-models","display_name":"quantized models","score":0.5320000052452087},{"id":"https://openalex.org/keywords/model-parameter-scaling","display_name":"model parameter scaling","score":0.44999998807907104},{"id":"https://openalex.org/keywords/llama","display_name":"llamas","score":0.5600000023841858},{"id":"https://openalex.org/keywords/qwen","display_name":"Qwen","score":0.7369999885559082},{"id":"https://openalex.org/keywords/leaderboards","display_name":"leaderboards","score":0.5910000205039978}],"concepts":[{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.682200014591217},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.6797000169754028},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.6557999849319458},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5641000270843506},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.45910000801086426},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43939998745918274},{"id":"https://openalex.org/C83725634","wikidata":"https://www.wikidata.org/wiki/Q7268699","display_name":"Qualitative reasoning","level":2,"score":0.4287000000476837},{"id":"https://openalex.org/C2777508537","wikidata":"https://www.wikidata.org/wiki/Q7936620","display_name":"Visual reasoning","level":2,"score":0.40209999680519104},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.3646000027656555},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3643999993801117},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.3637999892234802},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.362199991941452},{"id":"https://openalex.org/C2780898871","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Performance metric","level":2,"score":0.35989999771118164},{"id":"https://openalex.org/C171268870","wikidata":"https://www.wikidata.org/wiki/Q1486676","display_name":"GRASP","level":2,"score":0.3481999933719635},{"id":"https://openalex.org/C89288958","wikidata":"https://www.wikidata.org/wiki/Q7301504","display_name":"Reasoning system","level":2,"score":0.3361000120639801},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.32330000400543213},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.32179999351501465},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.31869998574256897},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.3149999976158142},{"id":"https://openalex.org/C2780339515","wikidata":"https://www.wikidata.org/wiki/Q3074698","display_name":"Arrow","level":2,"score":0.3070000112056732},{"id":"https://openalex.org/C76969082","wikidata":"https://www.wikidata.org/wiki/Q486902","display_name":"Mathematical model","level":2,"score":0.3043999969959259},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.2892000079154968},{"id":"https://openalex.org/C36964233","wikidata":"https://www.wikidata.org/wiki/Q7920942","display_name":"Verbal reasoning","level":3,"score":0.2867000102996826},{"id":"https://openalex.org/C166052673","wikidata":"https://www.wikidata.org/wiki/Q83021","display_name":"Empirical evidence","level":2,"score":0.2797999978065491},{"id":"https://openalex.org/C133199616","wikidata":"https://www.wikidata.org/wiki/Q25386885","display_name":"Empirical modelling","level":2,"score":0.27079999446868896}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1285","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1285","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1285","pdf_url":"https://aclanthology.org/2026.findings-acl.1285.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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"},{"id":"pmh:oai:arXiv.org:2507.04023","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2507.04023","pdf_url":"https://arxiv.org/pdf/2507.04023","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2507.04023","doi":"https://doi.org/10.48550/arxiv.2507.04023","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2507.04023","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,"listed_in":[],"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"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1285","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1285","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1285","pdf_url":"https://aclanthology.org/2026.findings-acl.1285.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G7160159671","display_name":"CAREER: Automated, Quality, and Trustworthy Scientific Information Extraction from Massive Text Data","funder_award_id":"2442253","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320307791","display_name":"Cisco Systems","ror":"https://ror.org/03yt1ez60"},{"id":"https://openalex.org/F4320309480","display_name":"Nvidia","ror":"https://ror.org/03jdj4y14"},{"id":"https://openalex.org/F4320312143","display_name":"National Centre for Supercomputing Applications","ror":"https://ror.org/03r10zj06"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4415314084.pdf","grobid_xml":"https://content.openalex.org/works/W4415314084.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4404782624","https://openalex.org/W4400612109","https://openalex.org/W4417299031","https://openalex.org/W7148793925","https://openalex.org/W4407359594","https://openalex.org/W7123300570","https://openalex.org/W4403885161","https://openalex.org/W4399554582","https://openalex.org/W4416036511","https://openalex.org/W7163370577"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2,135,172],"(LLMs)":[3],"achieve":[4],"impressive":[5],"performance":[6,121],"on":[7,14,44,122],"complex":[8,123],"mathematical":[9,196],"benchmarks":[10,124],"yet":[11],"sometimes":[12,141],"fail":[13],"basic":[15,83,130],"math":[16,84,131],"reasoning":[17,40,96,102,134,166,183,191],"while":[18,140],"generating":[19],"unnecessarily":[20],"verbose":[21],"responses.In":[22],"this":[23],"paper,":[24],"we":[25,52,57,73,86,104],"present":[26],"LLMTHINKBENCH,":[27],"a":[28,62,88],"systematic":[29],"benchmark":[30],"and":[31,50,67,97,112,145],"comprehensive":[32],"empirical":[33,90],"study":[34,91],"to":[35,129,156],"evaluate":[36],"the":[37,45,54,59,159,187],"efficiency":[38],"of":[39],"in":[41,192],"LLMs,":[42,94],"focusing":[43],"fundamental":[46],"tradeoff":[47],"between":[48],"accuracy":[49,66,144,175],"overthinking.First,":[51],"formalize":[53],"accuracy-verbosity":[55],"tradeoff.Second,":[56],"introduce":[58],"Overthinking":[60],"Score,":[61],"harmonic-mean":[63],"metric":[64],"combining":[65],"token-efficiency":[68],"for":[69,115],"holistic":[70],"model":[71,120],"evaluation.Third,":[72],"establish":[74],"an":[75,108],"evaluation":[76],"protocol":[77],"with":[78,164],"dynamically-generated":[79],"data":[80],"across":[81,100],"14":[82],"tasks.Fourth,":[85],"conduct":[87],"large-scale":[89],"evaluating":[92],"53":[93],"including":[95],"quantized":[98],"variants":[99],"different":[101],"budgets.Fifth,":[103],"release":[105],"LLM-THINKBENCH":[106],"as":[107],"open-source":[109],"Python":[110],"package":[111],"public":[113],"leaderboard":[114],"reproducibility.Our":[116],"findings":[117,185],"reveal:":[118],"1)":[119],"does":[125],"not":[126],"translate":[127],"directly":[128],"reasoning;":[132],"2)":[133],"generate":[136],"\u223c18\u00d7":[137],"more":[138],"tokens":[139,150],"achieving":[142],"lower":[143],"exhibit":[146],"catastrophic":[147],"collapse":[148],"when":[149],"are":[151],"constrained,":[152],"dropping":[153],"by":[154],"up":[155],"\u223c36%;":[157],"3)":[158],"accuracyverbosity":[160],"relationship":[161],"is":[162],"non-monotonic":[163],"extended":[165],"budgets":[167],"yielding":[168],"diminishing":[169],"returns":[170],"(GPT-5/o-series":[171],"show":[173],"zero":[174],"gain":[176],"from":[177],"low":[178],"\u2192":[179,181],"medium":[180],"high":[182],"effort).Our":[184],"challenge":[186],"assumption":[188],"that":[189],"longer":[190],"LLMs":[193],"necessarily":[194],"improves":[195],"reasoning.0":[197],".5":[198,297,300],"B":[199,202,204,206,208,210,212,214,216,219,222,225,228,250,253,255,257,260,263,298,301,303,305,308,311,314],"0":[200,224,232,248,266,296,317],".6":[201,249],"1":[203,205,207,217,251,258,299,306],"3":[209,220,261,302,309],"4":[211,218,254,259,307],"7":[213,223,226,304,312],"8":[215,256],"2":[221,227,262,310,313],"Model":[229,241,264,315],"Parameters":[230,265,316],"(B)":[231],"20":[233,267,318],"40":[234,268,319],"60":[235,269,320],"80":[236,270,321],"100":[237,271,322],"Accuracy":[238,272,323],"(%)":[239,273,324],"(a)":[240],"Family":[242],"Scaling":[243],"Comparison":[244],"Llama":[245],"Qwen2.5":[246],"Qwen3":[247],".7":[252],"(b)":[274],"Token":[275,279],"Budget":[276,280],"Impact":[277,327],"Full":[278],"Constrained":[281],"(1024":[282],"tokens)":[283],"50.0":[284],"70.2":[285],"81.9":[286],"82.1":[287],"86.5":[288],"84.1":[289],"27.0":[290],"30.7":[291],"31.7":[292],"28.4":[293],"35.2":[294],"30.2":[295],"(c)":[325],"Quantization":[326],"FP16":[328],"(Original)":[329],"GPTQ":[330,332],"8-bit":[331],"4-bit":[333]},"counts_by_year":[],"updated_date":"2026-10-26T11:51:55.731654","created_date":"2025-10-18T00:00:00"}