{"id":496,"date":"2026-10-07T13:14:53","date_gmt":"2026-10-07T13:14:53","guid":{"rendered":"https:\/\/bugleblast.com\/?p=496"},"modified":"2026-10-07T13:15:10","modified_gmt":"2026-10-07T13:15:10","slug":"openai-722-ai-math-manuscripts-github-lean","status":"publish","type":"post","link":"https:\/\/bugleblast.com\/?p=496","title":{"rendered":"OPENAI DUMPS 722 AI-WRITTEN MATH PAPERS ON GITHUB! | An unreleased model chewed through about 4,000 open problems at roughly three hours of ChatGPT Pro thinking per result \u2014 and OpenAI skipped the journals entirely while Fields medalists warn of a \u201csevere misalignment\u201d"},"content":{"rendered":"<p>Peer review, meet the firehose. OpenAI on Tuesday evening posted <strong>722 math manuscripts<\/strong>, organized into <strong>372 \u201cfamilies\u201d<\/strong> of related results, produced by an internal frontier model the public can\u2019t use yet, according to <a href=\"https:\/\/openai.com\/index\/sharing-ai-progress-in-mathematics\/\">OpenAI\u2019s announcement<\/a> and its <a href=\"https:\/\/github.com\/openai\/math\/tree\/main\">openai\/math GitHub repository<\/a>.<\/p>\n<p>The numbers: over the evaluation, the model was posed about <strong>4,000 problems<\/strong>. On average, each result used the equivalent of roughly <strong>three hours of ChatGPT Pro thinking<\/strong> compute, OpenAI says. <a href=\"https:\/\/the-decoder.com\/openai-dumps-372-ai-generated-math-proofs-on-github-telling-the-academic-world-to-keep-up\/\">The Decoder<\/a> reports nearly every result came from a single prompt to a single agent \u2014 a far cry from last month\u2019s Navier-Stokes claim, which took a swarm of up to 10,000 agents and millions of dollars in compute.<\/p>\n<p>What\u2019s in the pile: work across fields including number theory and algebraic geometry, and results the Wall Street Journal says touch three of the remaining Millennium Prize Problems, among them the <strong>Riemann hypothesis<\/strong>, per <a href=\"https:\/\/qz.com\/openai-math-results-github-millennium-prize-100726\">Quartz<\/a>. Many \u2014 but not all \u2014 of the papers come with <strong>Lean<\/strong> formalizations, computer-checkable proofs, and the repository itself warns the collection \u201cincludes results at different stages of verification.\u201d<\/p>\n<p>The bold move is where it lives: GitHub, not journals. OpenAI says it consulted the Institute for Advanced Study\u2019s Advisory Group on Mathematics and AI \u2014 but Quartz reports it broke with one request, that labs stop testing frontier math problems on proprietary models. OpenAI says it will keep doing that.<\/p>\n<p>Mathematicians aren\u2019t all cheering. In an open letter titled \u201cA Severe Misalignment of AI in Mathematics,\u201d 25 Fields Medal winners warned that mass-producing true statements could destroy fertile ground rather than create new ideas, The Decoder reports. OpenAI says it will fund workshops on AI-produced results and is working to \u201cresponsibly release\u201d the model. The humans now have homework \u2014 722 papers of it.<\/p>\n<p><strong>Sources:<\/strong> <a href=\"https:\/\/openai.com\/index\/sharing-ai-progress-in-mathematics\/\">OpenAI<\/a> \u00b7 <a href=\"https:\/\/github.com\/openai\/math\/tree\/main\">GitHub: openai\/math<\/a> \u00b7 <a href=\"https:\/\/the-decoder.com\/openai-dumps-372-ai-generated-math-proofs-on-github-telling-the-academic-world-to-keep-up\/\">The Decoder<\/a> \u00b7 <a href=\"https:\/\/qz.com\/openai-math-results-github-millennium-prize-100726\">Quartz<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>An unreleased model chewed through about 4,000 open problems at roughly three hours of ChatGPT Pro thinking per result \u2014 and OpenAI skipped the journals entirely while Fields medalists warn of a \u201csevere misalignment\u201d<\/p>\n","protected":false},"author":1,"featured_media":497,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_post_was_ever_published":false},"categories":[1],"tags":[],"class_list":["post-496","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-uncategorized"],"jetpack_sharing_enabled":true,"jetpack_featured_media_url":"https:\/\/bugleblast.com\/wp-content\/uploads\/2026\/10\/bb-openai-math-dump-1007.jpg","_links":{"self":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/posts\/496","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=496"}],"version-history":[{"count":1,"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/posts\/496\/revisions"}],"predecessor-version":[{"id":498,"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/posts\/496\/revisions\/498"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=\/wp\/v2\/media\/497"}],"wp:attachment":[{"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=496"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=496"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/bugleblast.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=496"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}