{"id":58131,"date":"2026-09-18T09:41:16","date_gmt":"2026-09-18T01:41:16","guid":{"rendered":"https:\/\/www.1ai.net\/?p=58131"},"modified":"2026-09-18T09:41:16","modified_gmt":"2026-09-18T01:41:16","slug":"%e8%b0%b7%e6%ad%8c%e6%9c%80%e5%bc%ba%e6%95%b0%e5%ad%a6%e6%8e%a8%e7%90%86-ai-%e6%a8%a1%e5%9e%8b%e6%9b%9d%e5%85%89%ef%bc%9a100-%e4%b8%87%e8%af%8d%e5%85%83%e4%b8%8a%e4%b8%8b%e6%96%87%ef%bc%8c%e4%b8%93","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/58131.html","title":{"rendered":"GOOGLE'S STRONGEST MATHEMATICAL REASONING AI MODEL EXPOSED: 1 MILLION WORDS IN CONTEXT, SPECIALIZING IN MATH PUZZLES"},"content":{"rendered":"<p>On September 18, the source @lyraxana posted tweets on platform X on September 16th<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e8%b0%b7%e6%ad%8c\" title=\"[View articles tagged with [Google]]\" target=\"_blank\" >Google<\/a>Based on the DeepThink V3 model<strong>Build an internal experiment codenamed Mathematica <a href=\"https:\/\/www.1ai.net\/en\/tag\/ai%e6%a8%a1%e5%9e%8b\" title=\"[View articles tagged with [AI models]]\" target=\"_blank\" >AI Models<\/a>, WHICH IS ESPECIALLY OPTIMIZED FOR SOLVING COMPLEX SYMBOL PROBLEMS, MAINLY IN HEAVY COMPUTING, AND IS EXPECTED TO BE THE BEST MATHEMATICAL REASONING AI MODEL OF GOOGLE\u3002<\/strong><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-58132\" title=\"b3618467jljdzl0008fd000ht00bfp\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2026\/09\/b3618467j00tljdzl008fd000ht00bfp.jpg\" alt=\"b3618467jljdzl0008fd000ht00bfp\" width=\"641\" height=\"411\" \/><\/p>\n<p>API data show that the model's internal identifier is \u201cmodels \/ deepthink-mathematica-tf-raw-thoughts\u201d and supports context windows up to 1 million characters (token), with an output limit of 65,536\u3002<\/p>\n<p>Note: The context window refers to the total number of text units that can be read while the model is being processed; the output ceiling refers to the number of text units that can be generated by the model's single response\u3002<\/p>\n<p>The leak configuration also labels \"UNSTABLE_EXPERIMETAL\". This status name indicates that the version is an unstable experimental version. The model is accompanied by the \u201cTeamfood\u201d label, which is the term Google uses for internal employee testing, meaning that the service is within the scope of internal testing\u3002<\/p>\n<p>TestingCatalog points out that Google has previously launched the Gemini DeepThink IMO model for high-level mathematics competitions and the Gemini 3.8 Live and Live Extened Thinking on September 15\u3002<\/p>","protected":false},"excerpt":{"rendered":"<p>On September 18, the source @lyraxana posted a tweet on platform X on September 16, in which Google is based on the DeepThink V3 model, building an experimental AI model with the internal code name Mathematica, which seeks special optimization, mainly for heavy computing and complex symbol questions, and is expected to be the best mathematical reasoning AI model for Google. API data show that the model's internal identifier is \u201cmodels \/ deepthink-mathematica-tf-raw-thoughts\u201d and supports context windows up to 1 million characters (token), with an output limit of 65,536 words. Notes:<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[167,281],"collection":[],"class_list":["post-58131","post","type-post","status-publish","format-standard","hentry","category-news","tag-ai","tag-281"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/58131","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=58131"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/58131\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=58131"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=58131"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=58131"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=58131"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}