[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82934-en":3,"doc-seo-82934-105":29,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82934,8796095461610,"Oliver","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","OptiAgent End-to-End Optimization Modeling via Multi-Agent Iterative Refinement","OptiAgent is a multi-agent framework that converts a natural-language description of an Operations Research problem into a mathematically precise, solver-ready formulation and executable code. The system centers on the modeling stage by using specialized agents to extract decision variables, constraints, and objective structure, then apply iterative self-correction. A multi-loop validation design with four feedback mechanisms targets distinct failure modes including misinterpretation, structural defects, mathematical inconsistencies, validation failures, and code errors. Evaluations show state-of-the-art performance on multiple LP, MILP, and nonlinear benchmarks with strong competitiveness elsewhere.","arXiv :2607 .05346v 1 [ cs .AI] 6 Jul 2026  \nOptiAgent: End-to-End Optimization Modeling via Multi-Agent Iterative Refinement  \nAdriana Laurindo Monteiro∗ Nayse Fagundes∗ Gabriel Mattos Langeloh  \nGustavo de Oliveira Kanno Priscila Louise Aguirre  \nThiago Costa Rizuti da Rocha†  \nVictor Leme Beltran  \nInstituto de Ciência e Tecnologia do Itaú, Brasil  \n[mlaurindodrica@gmail.com](mlaurindodrica@gmail.com), [naysesfagundes@gmail.com](naysesfagundes@gmail.com)  \n[gabriel.langeloh@itau-unibanco.com.br](gabriel.langeloh@itau-unibanco.com.br), [gustavo.kanno@itau-unibanco.com.br](gustavo.kanno@itau-unibanco.com.br)[priscila.aguirre@itau-unibanco.com.br](priscila.aguirre@itau-unibanco.com.br), [thiago.rizuti-rocha@itau-unibanco.com.br](thiago.rizuti-rocha@itau-unibanco.com.br)  \n[victor.beltran@itau-unibanco.com.br](victor.beltran@itau-unibanco.com.br)  \nAbstract  \nWe propose OptiAgent, a multi-agent framework that, given a natural language description of an Operations Research problem, is able to output a solver-ready mathematical formulation as well as executable code. Our architecture prioritizes the mathematical modeling step, where dedicated agents extract structures, such as decision variables and constraints, enabling iterative self-correction. We introduce a novel multi-loop validation architecture with four specialized feedback mechanisms, each targeting a distinct failure mode such as misinterpretation, structural defects, mathematical inconsistencies, validation failures, and code errors. Alongside accuracy, our modular design improves the process of solving optimization problems by improving transparency, as each agent exposes its reasoning and feedback, making the full modeling process auditable. Our framework achieves state-of-the-art performance on 3 out of 4 benchmarks across LP, MILP, and Nonlinear Programming tasks, while remaining highly competitive on the remaining dataset.  \nKeywords: Multi-Agent; Operations Research; Optimization Modeling; Trustworthy AI; Decision Support Systems.  \n1 Introduction  \nThe increasing performance of Large Language Models (LLMs), whether it is caused by a higher number of parameters, or larger context windows or better reasoning capabilities has been accompanied with harder and even more complex tasks to Artificial Intelligence  \n∗ Equal contribution as first authors.†Corresponding author.  \n(AI) . One example in this context is the use of LLMs to model optimization problems described in Natural Language (NL) .  \nThe ultimate goal is to use the reasoning capabilities of LLMs to identify the mathematical structure behind a NL description of an optimization problem. The model has to extract abstract entities of the problem: the decision variables, the feasible set structured by constraint functions and the objective function. The process of formulating a precise mathematical model, that is, to translate the natural language description, is highly costly: one has to be an Operations Research (OR) expert. This poses a huge barrier to industry applications, since solvers require precisely formulated mathematical models.  \nIn this context, the use of LLMs as solvers or modeling assistants is a valuable tool as it can reduce the time and cost of the optimization modeling process. The research field of OR is concerned with a huge variety of classical optimization problems, ranging from Linear Programming (LP), Mixed-Integer Linear Programming (MILP), nonlinear, stochastic, convex and nonconvex problems to combinatorial problems. Although previous research has established the use of LLMs for formulating and solving optimization problems described by NL, a few challenges remain unsolved.  \nIn particular, moving from research demonstrations to practical tools exposes several methodological and usability challenges. As discussed in (Xiao et al. , 2025), we can cite the cost of crafting a highly specialized dataset and the need of better evaluation methods, usually measured in solution accuracy (","cbCaitayUYkdoNNX","https://ap.wps.com/l/cbCaitayUYkdoNNX","pdf",591779,1,16,"English","en",105,"# Abstract\n# Introduction\n## Contributions","[{\"question\":\"What does OptiAgent produce from a natural-language description?\",\"answer\":\"It outputs a solver-ready mathematical formulation and executable code, derived from the described Operations Research problem.\"},{\"question\":\"How does OptiAgent improve reliability during modeling?\",\"answer\":\"It uses a modeling-focused multi-agent setup with iterative self-correction and a multi-loop validation architecture with four feedback mechanisms targeting different failure modes.\"},{\"question\":\"What optimization problem types does OptiAgent support and how does it perform?\",\"answer\":\"It addresses LP, MILP, and nonlinear programming tasks, achieving state-of-the-art results on three of four benchmarks and remaining highly competitive on the remaining dataset.\"}]",1784184098,40,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"optiagent-end-to-end-optimization-modeling-via-multi-agent-iterative-refinement","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/optiagent-end-to-end-optimization-modeling-via-multi-agent-iterative-refinement/82934/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What does OptiAgent produce from a natural-language description?","Question",{"text":74,"@type":75},"It outputs a solver-ready mathematical formulation and executable code, derived from the described Operations Research problem.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does OptiAgent improve reliability during modeling?",{"text":79,"@type":75},"It uses a modeling-focused multi-agent setup with iterative self-correction and a multi-loop validation architecture with four feedback mechanisms targeting different failure modes.",{"name":81,"@type":72,"acceptedAnswer":82},"What optimization problem types does OptiAgent support and how does it perform?",{"text":83,"@type":75},"It addresses LP, MILP, and nonlinear programming tasks, achieving state-of-the-art results on three of four benchmarks and remaining highly competitive on the remaining dataset.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,118,121,126,129,133],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":28,"slug":117},7,"Healthcare","healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":119,"slug":120},30,"research-report",{"id":122,"doc_module":4,"doc_module_name":45,"category_name":123,"show_sort_weight":124,"slug":125},9,"Religion & Spirituality",20,"religion-spirituality",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":124,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]