{"id":"https://openalex.org/W7163216513","doi":"https://doi.org/10.48550/arxiv.2606.02031","title":"OpenWebRL: Demystifying Online Multi-turn Reinforcement Learning for Visual Web Agents","display_name":"OpenWebRL: Demystifying Online Multi-turn Reinforcement Learning for Visual Web Agents","publication_year":2026,"publication_date":"2026-06-01","ids":{"openalex":"https://openalex.org/W7163216513","doi":"https://doi.org/10.48550/arxiv.2606.02031"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.02031","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.02031","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":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.2606.02031","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137692706","display_name":"Rui Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Rui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137704722","display_name":"Qianhui Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Qianhui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137615163","display_name":"Yuxi Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Yuxi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137667988","display_name":"Hao Bai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bai, Hao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137628862","display_name":"Wenlin Yao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yao, Wenlin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137693219","display_name":"Hao Cheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cheng, Hao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137637203","display_name":"Baolin Peng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng, Baolin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137709402","display_name":"Huan Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Huan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137637805","display_name":"Tong Zhang","orcid":"https://orcid.org/0000-0002-0557-7090"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Tong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137657287","display_name":"Jianfeng Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Jianfeng","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9115999937057495,"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"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9115999937057495,"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":0.011599999852478504,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.00839999970048666,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.7042999863624573},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.6549000144004822},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5521000027656555},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5098999738693237},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5002999901771545},{"id":"https://openalex.org/keywords/initialization","display_name":"Initialization","score":0.4553999900817871},{"id":"https://openalex.org/keywords/web-application","display_name":"Web application","score":0.4309000074863434},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.33899998664855957}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7807000279426575},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7042999863624573},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6549000144004822},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5521000027656555},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5098999738693237},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5002999901771545},{"id":"https://openalex.org/C136764020","wikidata":"https://www.wikidata.org/wiki/Q466","display_name":"World Wide Web","level":1,"score":0.46480000019073486},{"id":"https://openalex.org/C114466953","wikidata":"https://www.wikidata.org/wiki/Q6034165","display_name":"Initialization","level":2,"score":0.4553999900817871},{"id":"https://openalex.org/C118643609","wikidata":"https://www.wikidata.org/wiki/Q189210","display_name":"Web application","level":2,"score":0.4309000074863434},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.39419999718666077},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38040000200271606},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.33899998664855957},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.33820000290870667},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3314000070095062},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.3312999904155731},{"id":"https://openalex.org/C130436687","wikidata":"https://www.wikidata.org/wiki/Q7978591","display_name":"Web modeling","level":3,"score":0.31470000743865967},{"id":"https://openalex.org/C11392498","wikidata":"https://www.wikidata.org/wiki/Q11288","display_name":"Web server","level":3,"score":0.3003000020980835},{"id":"https://openalex.org/C110875604","wikidata":"https://www.wikidata.org/wiki/Q75","display_name":"The Internet","level":2,"score":0.2732999920845032},{"id":"https://openalex.org/C35578498","wikidata":"https://www.wikidata.org/wiki/Q193424","display_name":"Web service","level":2,"score":0.2689000070095062},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.2669000029563904},{"id":"https://openalex.org/C61096286","wikidata":"https://www.wikidata.org/wiki/Q7978592","display_name":"Web navigation","level":3,"score":0.2612999975681305},{"id":"https://openalex.org/C136389625","wikidata":"https://www.wikidata.org/wiki/Q334384","display_name":"Supervised learning","level":3,"score":0.2563000023365021}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.02031","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.02031","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":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.2606.02031","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.02031","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/9","score":0.6578570008277893,"display_name":"Industry, innovation and infrastructure"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Building":[0],"capable":[1],"visual":[2,80,101,212],"web":[3,40,81,102,213,237],"agents":[4,28,82,103,177],"requires":[5],"long-horizon":[6],"reasoning,":[7],"precise":[8],"grounding,":[9],"and":[10,55,129,158,170,183,192,215,234,246],"robust":[11],"interaction":[12],"with":[13,104,186],"dynamic":[14],"real-world":[15],"websites.":[16,110],"Despite":[17],"rapid":[18],"progress,":[19],"the":[20,62,113,147,202],"strongest":[21],"systems":[22,188],"remain":[23],"largely":[24,88],"proprietary,":[25],"while":[26],"open":[27,65,97,176,236],"still":[29],"depend":[30],"heavily":[31],"on":[32,84,108,149,168,172],"supervised":[33,121],"post-training":[34],"over":[35],"large":[36],"collections":[37],"of":[38,61,146,178],"curated":[39],"trajectories.":[41],"This":[42],"dependence":[43],"creates":[44],"a":[45,142,226],"major":[46],"scalability":[47],"bottleneck:":[48],"high-quality":[49],"demonstrations":[50],"are":[51],"expensive":[52],"to":[53,248],"collect,":[54],"static":[56],"datasets":[57],"offer":[58],"limited":[59],"coverage":[60],"diverse,":[63],"ever-changing":[64],"web.":[66],"Although":[67],"online":[68,105,208],"RL":[69,107,161,209,218],"has":[70],"shown":[71],"promise":[72],"for":[73,78,99,211],"text-based":[74],"agents,":[75,214],"its":[76],"potential":[77],"training":[79,100,115,162,243],"directly":[83],"live":[85],"websites":[86],"remains":[87],"underexplored.":[89],"In":[90],"this":[91,135],"paper,":[92],"we":[93,137,199],"introduce":[94],"OpenWebRL,":[95],"an":[96],"framework":[98],"multi-turn":[106,131],"real":[109],"OpenWebRL":[111],"covers":[112],"full":[114],"pipeline,":[116],"including":[117,189],"scalable":[118],"live-browser":[119],"infrastructure,":[120],"initialization,":[122],"multimodal":[123],"context":[124],"management,":[125],"trajectory-level":[126],"success":[127,167],"judging,":[128],"efficient":[130],"policy":[132],"optimization.":[133],"Using":[134],"framework,":[136],"train":[138],"OpenWebRL-4B,":[139],"which":[140],"establishes":[141],"new":[143],"open-source":[144],"state":[145],"art":[148],"challenging":[150],"live-web":[151],"benchmarks.":[152],"With":[153],"only":[154],"0.4K":[155],"initialization":[156],"trajectories":[157],"2.2K":[159],"open-ended":[160],"tasks,":[163],"OpenWebRL-4B":[164],"achieves":[165],"67.0%":[166],"Online-Mind2Web":[169],"64.0%":[171],"DeepShop,":[173],"outperforming":[174],"prior":[175],"similar":[179],"or":[180],"larger":[181],"scale":[182],"remaining":[184],"competitive":[185],"proprietary":[187],"OpenAI":[190],"CUA":[191],"Gemini":[193],"CUA.":[194],"Beyond":[195],"strong":[196],"benchmark":[197],"performance,":[198],"systematically":[200],"study":[201],"key":[203],"design":[204],"choices":[205],"that":[206],"make":[207],"effective":[210],"analyze":[216],"how":[217],"improves":[219],"agentic":[220],"reasoning.":[221],"Overall,":[222],"our":[223,242],"work":[224],"offers":[225],"practical":[227],"path":[228],"toward":[229],"building":[230],"more":[231],"capable,":[232],"reproducible,":[233],"cost-efficient":[235],"agents.":[238],"We":[239],"will":[240],"release":[241],"data,":[244],"models,":[245],"code":[247],"support":[249],"future":[250],"research.":[251]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-03T00:00:00"}