{"id":"https://openalex.org/W7166676514","doi":"https://doi.org/10.48550/arxiv.2606.29150","title":"Flow Reasoning Models: Scaling Reasoning Through Iterative Self-Refinement","display_name":"Flow Reasoning Models: Scaling Reasoning Through Iterative Self-Refinement","publication_year":2026,"publication_date":"2026-06-28","ids":{"openalex":"https://openalex.org/W7166676514","doi":"https://doi.org/10.48550/arxiv.2606.29150"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.29150","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.29150","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.2606.29150","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139712823","display_name":"Alec Helbling","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Helbling, Alec","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139675534","display_name":"Andrey Bryutkin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bryutkin, Andrey","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5031120336","display_name":"Mauro Martino","orcid":"https://orcid.org/0009-0009-2510-6120"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Martino, Mauro","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078338477","display_name":"Nima Dehmamy","orcid":"https://orcid.org/0000-0003-1617-5502"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dehmamy, Nima","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5084459196","display_name":"Hendrik Strobelt","orcid":"https://orcid.org/0000-0002-8995-1683"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Strobelt, Hendrik","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":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/T11574","display_name":"Artificial Intelligence in Games","score":0.1298000067472458,"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/T11574","display_name":"Artificial Intelligence in Games","score":0.1298000067472458,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.0940999984741211,"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"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.08209999650716782,"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/flow","display_name":"Flow (mathematics)","score":0.5900999903678894},{"id":"https://openalex.org/keywords/base","display_name":"Base (topology)","score":0.5407999753952026},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.5194000005722046},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5135999917984009},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.4799000024795532},{"id":"https://openalex.org/keywords/channel","display_name":"Channel (broadcasting)","score":0.3806999921798706},{"id":"https://openalex.org/keywords/recipe","display_name":"Recipe","score":0.3517000079154968},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.3386000096797943}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.677299976348877},{"id":"https://openalex.org/C38349280","wikidata":"https://www.wikidata.org/wiki/Q1434290","display_name":"Flow (mathematics)","level":2,"score":0.5900999903678894},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.5407999753952026},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.5194000005722046},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5135999917984009},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4927999973297119},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.4799000024795532},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.3806999921798706},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.35839998722076416},{"id":"https://openalex.org/C2778671685","wikidata":"https://www.wikidata.org/wiki/Q219239","display_name":"Recipe","level":2,"score":0.3517000079154968},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.34119999408721924},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.3386000096797943},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.3357999920845032},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3343999981880188},{"id":"https://openalex.org/C61445026","wikidata":"https://www.wikidata.org/wiki/Q217608","display_name":"Fixed point","level":2,"score":0.33379998803138733},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.32850000262260437},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.2971000075340271},{"id":"https://openalex.org/C159718280","wikidata":"https://www.wikidata.org/wiki/Q5526353","display_name":"Gas meter prover","level":3,"score":0.29440000653266907},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.28189998865127563},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.2687000036239624},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.26809999346733093},{"id":"https://openalex.org/C86827895","wikidata":"https://www.wikidata.org/wiki/Q7098582","display_name":"Opportunistic reasoning","level":4,"score":0.26019999384880066},{"id":"https://openalex.org/C157469704","wikidata":"https://www.wikidata.org/wiki/Q2585642","display_name":"Maximum flow problem","level":2,"score":0.25290000438690186}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.29150","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.29150","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.2606.29150","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.29150","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":[{"score":0.43791642785072327,"id":"https://metadata.un.org/sdg/1","display_name":"No poverty"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Discrete":[0],"flow":[1,53,65,165,239],"models":[2,66,166,186,240],"have":[3],"recently":[4],"shown":[5],"promising":[6],"performance":[7],"on":[8,114,133,209],"few-step":[9],"text":[10],"generation;":[11],"however,":[12,139],"when":[13,88],"naively":[14],"applied":[15],"to":[16,29,86,123,156,187],"structured":[17,50],"reasoning":[18,51],"tasks":[19],"such":[20],"as":[21,69],"Sudoku":[22,36,210],"and":[23,46,90,101,117,171],"Zebra":[24,118],"puzzles,":[25],"they":[26],"converge":[27],"confidently":[28],"incorrect":[30,147],"answers":[31],"(solving":[32],"only":[33],"$\\sim$36%":[34],"of":[35,81,144],"puzzles).":[37],"We":[38,55,150],"introduce":[39],"Flow":[40],"Reasoning":[41],"Models":[42],"(FRMs),":[43],"a":[44,77,94,141,153,168],"training":[45,132,154],"test-time-scaling":[47,95],"framework":[48],"for":[49,228],"with":[52,167,234],"models.":[54],"make":[56],"the":[57,82,158,201,220,229],"observation":[58],"that,":[59],"despite":[60],"their":[61,70,179,189],"poor":[62],"solve":[63,112,241],"rate,":[64],"can":[67],"act":[68],"own":[71,180,190],"verifiers.":[72],"A":[73],"correct":[74],"answer":[75],"is":[76],"stable":[78],"fixed":[79],"point":[80],"denoising":[83],"dynamics,":[84],"returning":[85],"itself":[87],"re-noised":[89],"re-solved.":[91],"This":[92,120,136],"enables":[93],"paradigm:":[96],"propose":[97],"many":[98],"candidate":[99,148],"solutions":[100],"keep":[102],"those":[103],"that":[104,134],"are":[105],"dynamically":[106],"stable,":[107],"which":[108],"alone":[109],"reaches":[110],"high":[111],"rates":[113],"Sudoku-Shah":[115],"(~$100\\%$)":[116],"($95.9\\%$).":[119],"even":[121],"generalizes":[122],"harder":[124],"out-of-distribution":[125,243],"puzzles":[126,244],"like":[127],"Sudoku-Extreme":[128],"($96.1\\%$),":[129],"without":[130],"ever":[131],"distribution.":[135],"pure":[137],"search,":[138],"wastes":[140],"great":[142],"deal":[143],"computation":[145],"generating":[146],"solutions.":[149],"therefore":[151],"design":[152],"recipe":[155],"improve":[157,200],"base":[159,202],"model's":[160,203],"efficiency.":[161],"First,":[162],"we":[163,184,225],"train":[164,185],"self-conditioning":[169],"channel":[170],"close":[172],"it":[173,206],"at":[174],"inference,":[175],"letting":[176,205],"them":[177],"refine":[178],"past":[181],"predictions.":[182],"Second,":[183],"avoid":[188],"failed":[191],"generations":[192],"using":[193],"direct":[194],"preference":[195],"optimization.":[196],"These":[197],"changes":[198],"substantially":[199],"efficiency,":[204],"reach":[207],"$99.2\\%$":[208],"in":[211],"just":[212],"$7$":[213],"forward":[214],"passes,":[215],"over":[216],"$8\\times$":[217],"fewer":[218],"than":[219],"strongest":[221],"matched":[222],"masked-diffusion":[223],"baseline":[224],"compare":[226],"needs":[227],"same":[230],"accuracy.":[231],"When":[232],"combined":[233],"test-time":[235],"scaling,":[236],"this":[237],"lets":[238],"hard":[242],"(e.g.":[245],"Sudoku-Extreme)":[246],"far":[247],"more":[248],"efficiently.":[249]},"counts_by_year":[],"updated_date":"2026-07-01T06:29:00.853634","created_date":"2026-07-01T00:00:00"}