{"id":1174706,"date":"2026-06-04T12:26:58","date_gmt":"2026-06-04T19:26:58","guid":{"rendered":"https:\/\/find.codeghost.online\/en-us\/research\/publication\/frontieror-benchmarking-llms-capacity-for-efficient-algorithm-design-in-large-scale-optimization\/"},"modified":"2026-06-08T17:05:25","modified_gmt":"2026-06-09T00:05:25","slug":"frontieror-benchmarking-llms-capacity-for-efficient-algorithm-design-in-large-scale-optimization","status":"publish","type":"msr-research-item","link":"https:\/\/find.codeghost.online\/en-us\/research\/publication\/frontieror-benchmarking-llms-capacity-for-efficient-algorithm-design-in-large-scale-optimization\/","title":{"rendered":"FrontierOR: Benchmarking LLMs&#8217;Capacity for Efficient Algorithm Design in Large-Scale Optimization"},"content":{"rendered":"\n\n\n<p class=\"wp-block-paragraph\">Large language models (LLMs) are increasingly used for optimization modeling and solver-code generation, yet practical operations research and optimization problems often require a harder capability: designing scalable algorithms that exploit problem structure and outperform direct formulation-and-solve baselines. Existing benchmarks are limited to small or simplified examples far below real-world scale and complexity. We introduce FrontierOR, among the first benchmarks to systematically evaluate LLM-based efficient algorithm design for realistic large-scale optimization problems. FrontierOR includes 180 tasks derived from methodologically diverse papers published in top-tier operations research venues, each with standardized instances and a hidden, expert-verified evaluation suite. We evaluate seven LLMs spanning frontier, cost-effective, and open-source models both in one-shot and test-time evolution settings. The results reveal that frontier models still struggle to move from executable formulations to efficient optimization algorithms: the strongest one-shot model outperforms Gurobi in only 31% of cases in both solution quality and computational efficiency, and even strong coding agents with test-time evolution achieve only 50% on selected hard tasks. FrontierOR establishes a practical evaluation platform for LLM-based optimization algorithm design, which enables future LLMs and agents to be systematically tested on whether they can move beyond correct formulation toward a feasible, high-quality, and efficient algorithm. Code and data are publicly released at https:\/\/github.com\/Minw913\/FrontierOR.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Large language models (LLMs) are increasingly used for optimization modeling and solver-code generation, yet practical operations research and optimization problems often require a harder capability: designing scalable algorithms that exploit problem structure and outperform direct formulation-and-solve baselines. Existing benchmarks are limited to small or simplified examples far below real-world scale and complexity. We introduce FrontierOR, [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[{"type":"name","value":"Minwei Kong","user_id":0},{"type":"name","value":"Chonghe Jiang","user_id":0},{"type":"name","value":"Ao Qu","user_id":0},{"type":"name","value":"Wenbin Ouyang","user_id":0},{"type":"name","value":"Zhaoming Zeng","user_id":0},{"type":"name","value":"Xiaotong Guo","user_id":0},{"type":"name","value":"Zhekai Li","user_id":0},{"type":"name","value":"Junyi Li","user_id":0},{"type":"name","value":"Yingying Fan","user_id":0},{"type":"name","value":"Xinshou Zheng","user_id":0},{"type":"name","value":"Xibin Jing","user_id":0},{"type":"name","value":"Yikai Zhang","user_id":0},{"type":"name","value":"Zhiwei Liang","user_id":0},{"type":"name","value":"Seong-Hee Kim","user_id":0},{"type":"name","value":"Runqing Yang","user_id":0},{"type":"name","value":"Zijian Zhou","user_id":0},{"type":"name","value":"Sirui Li","user_id":0},{"type":"name","value":"Han Zheng","user_id":0},{"type":"name","value":"Wangyang Ying","user_id":0},{"type":"name","value":"Ou Zheng","user_id":0},{"type":"user_nicename","value":"Chong Wang","user_id":"35612"},{"type":"user_nicename","value":"Jing Zhao","user_id":"32307"},{"type":"name","value":"Hanzhang Qin","user_id":0},{"type":"user_nicename","value":"Cathy Wu","user_id":"38088"},{"type":"name","value":"Paul Liang","user_id":0},{"type":"name","value":"Jinhua Zhao","user_id":0},{"type":"name","value":"Hai Wang","user_id":0}],"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"arXiv","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"","msr_doi":"","msr_arxiv_id":"","msr_mag_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_release_tracker_id":"","msr_highlight_type":"","msr_date_display_format":"","msr_main_download_label":"","msr_external_link_label":"","msr_doi_label":"","msr_published_date":"2026-05-24","msr_startdate":"","msr_presentation_date":"","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_year":2026,"msr_month":5,"msr_day":24,"msr_microsoftintellectualproperty":false,"msr_pub_id":"","msr_publication_uploader":[{"type":"url","viewUrl":"false","id":false,"title":"https:\/\/arxiv.org\/abs\/2605.25246","label_id":243109,"label":0}],"msr_related_uploader":[],"msr_original_fields_of_study":[],"msr_s2_paper_id":"","msr_s2_pdf_url":"","msr_citation_count_updated":"","msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":0,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[{"provider":"s2","id":"4f1bf445a4fbed7adcb85e9619dd79fed8f41398"},{"provider":"arxiv","id":"2605.25246"}],"msr_hide_image_in_river":null,"footnotes":""},"msr-research-highlight":[],"research-area":[13556,13546],"msr-publication-type":[270373],"msr-publisher":[],"msr-publication-cta":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[246694,246691],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-1174706","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-artificial-intelligence","msr-research-area-computational-sciences-mathematics","msr-locale-en_us","msr-field-of-study-artificial-intelligence","msr-field-of-study-computer-science"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2026-05-24","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"arXiv","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":0,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"url","viewUrl":"false","id":"false","title":"https:\/\/arxiv.org\/abs\/2605.25246","label_id":"243109","label":0}],"msr_related_uploader":[],"msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"","msr_influential_citations":0,"msr_reference_count":0,"msr_arxiv_id":"","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[],"msr-author-ordering":[{"type":"name","value":"Minwei Kong","user_id":0,"rest_url":false},{"type":"name","value":"Chonghe Jiang","user_id":0,"rest_url":false},{"type":"name","value":"Ao Qu","user_id":0,"rest_url":false},{"type":"name","value":"Wenbin Ouyang","user_id":0,"rest_url":false},{"type":"name","value":"Zhaoming Zeng","user_id":0,"rest_url":false},{"type":"name","value":"Xiaotong Guo","user_id":0,"rest_url":false},{"type":"name","value":"Zhekai Li","user_id":0,"rest_url":false},{"type":"name","value":"Junyi Li","user_id":0,"rest_url":false},{"type":"name","value":"Yingying Fan","user_id":0,"rest_url":false},{"type":"name","value":"Xinshou Zheng","user_id":0,"rest_url":false},{"type":"name","value":"Xibin Jing","user_id":0,"rest_url":false},{"type":"name","value":"Yikai Zhang","user_id":0,"rest_url":false},{"type":"name","value":"Zhiwei Liang","user_id":0,"rest_url":false},{"type":"name","value":"Seong-Hee Kim","user_id":0,"rest_url":false},{"type":"name","value":"Runqing Yang","user_id":0,"rest_url":false},{"type":"name","value":"Zijian Zhou","user_id":0,"rest_url":false},{"type":"name","value":"Sirui Li","user_id":0,"rest_url":false},{"type":"name","value":"Han Zheng","user_id":0,"rest_url":false},{"type":"name","value":"Wangyang Ying","user_id":0,"rest_url":false},{"type":"name","value":"Ou Zheng","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Chong Wang","user_id":35612,"rest_url":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Chong Wang"},{"type":"user_nicename","value":"Jing Zhao","user_id":32307,"rest_url":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Jing Zhao"},{"type":"name","value":"Hanzhang Qin","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Cathy Wu","user_id":38088,"rest_url":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Cathy Wu"},{"type":"name","value":"Paul Liang","user_id":0,"rest_url":false},{"type":"name","value":"Jinhua Zhao","user_id":0,"rest_url":false},{"type":"name","value":"Hai Wang","user_id":0,"rest_url":false}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"misc","related_content":[],"_links":{"self":[{"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1174706","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1174706\/revisions"}],"predecessor-version":[{"id":1175098,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1174706\/revisions\/1175098"}],"wp:attachment":[{"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1174706"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=1174706"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1174706"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=1174706"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=1174706"},{"taxonomy":"msr-publication-cta","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-cta?post=1174706"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=1174706"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1174706"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1174706"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=1174706"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=1174706"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=1174706"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=1174706"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/find.codeghost.online\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=1174706"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}