{"id":27923,"date":"2025-01-29T15:06:26","date_gmt":"2025-01-29T07:06:26","guid":{"rendered":"https:\/\/www.1ai.net\/?p=27923"},"modified":"2025-01-29T15:06:26","modified_gmt":"2025-01-29T07:06:26","slug":"%e9%80%9a%e4%b9%89%e5%8d%83%e9%97%ae-qwen-2-5-max-%e8%b6%85%e5%a4%a7%e8%a7%84%e6%a8%a1-moe-%e6%a8%a1%e5%9e%8b%e5%8f%b7%e7%a7%b0%e4%bc%98%e4%ba%8e-deepseek-v3-%e7%ad%89%e7%ab%9e%e5%93%81%ef%bc%8c","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/27923.html","title":{"rendered":"The Qwen 2.5-Max hyperscale MoE model is claimed to be better than Deepseek V3 and other competitors, and has not been open-sourced for the time being."},"content":{"rendered":"<p>1 \u6708 29 \u65e5\u6d88\u606f\uff0c\u65b0\u5e74\u4e4b\u9645\uff0c<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e9%98%bf%e9%87%8c%e4%ba%91\" title=\"\u3010\u67e5\u770b\u542b\u6709[\u963f\u91cc\u4e91]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">\u963f\u91cc\u4e91<\/a>\u516c\u5e03\u4e86\u5176\u5168\u65b0\u7684<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e9%80%9a%e4%b9%89%e5%8d%83%e9%97%ae\" title=\"\u3010\u67e5\u770b\u542b\u6709[\u901a\u4e49\u5343\u95ee]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">\u901a\u4e49\u5343\u95ee<\/a> Qwen 2.5-Max \u8d85\u5927\u89c4\u6a21 <a href=\"https:\/\/www.1ai.net\/en\/tag\/moe%e6%a8%a1%e5%9e%8b\" title=\"\u3010\u67e5\u770b\u542b\u6709[MoE\u6a21\u578b]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">MoE\u6a21\u578b<\/a>\uff0c\u5927\u5bb6\u53ef\u4ee5\u901a\u8fc7 API \u7684\u65b9\u5f0f\u8fdb\u884c\u8bbf\u95ee\uff0c\u4e5f\u53ef\u4ee5\u767b\u5f55 Qwen Chat \u8fdb\u884c\u4f53\u9a8c\uff0c\u4f8b\u5982\u76f4\u63a5\u4e0e\u6a21\u578b\u5bf9\u8bdd\uff0c\u6216\u8005\u4f7f\u7528 artifacts\u3001\u641c\u7d22\u7b49\u529f\u80fd\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-27924\" title=\"55787359j00squ8yw00cpd000v900dap\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/01\/55787359j00squ8yw00cpd000v900dap.jpg\" alt=\"55787359j00squ8yw00cpd000v900dap\" width=\"1125\" height=\"478\" \/><\/p>\n<p>\u636e\u4ecb\u7ecd\uff0c\u901a\u4e49\u5343\u95ee Qwen 2.5-Max \u4f7f\u7528\u8d85\u8fc7 20 \u4e07\u4ebf token \u7684\u9884\u8bad\u7ec3\u6570\u636e\u53ca\u7cbe\u5fc3\u8bbe\u8ba1\u7684\u540e\u8bad\u7ec3\u65b9\u6848\u8fdb\u884c\u8bad\u7ec3\u3002<\/p>\n<p><strong>\u6027\u80fd<\/strong><\/p>\n<p>\u963f\u91cc\u4e91\u76f4\u63a5\u5bf9\u6bd4\u4e86\u6307\u4ee4\u6a21\u578b\u7684\u6027\u80fd\u8868\u73b0\uff08\u6ce8\uff1a\u6307\u4ee4\u6a21\u578b\u5373\u6211\u4eec\u5e73\u5e38\u4f7f\u7528\u7684\u53ef\u4ee5\u76f4\u63a5\u5bf9\u8bdd\u7684\u6a21\u578b\uff09\u3002\u5bf9\u6bd4\u5bf9\u8c61\u5305\u62ec DeepSeek V3\u3001GPT-4o \u548c Claude-3.5-Sonnet\uff0c\u7ed3\u679c\u5982\u4e0b\uff1a<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-27926\" title=\"9b12780fj00squ90t003od000v900hrp\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/01\/9b12780fj00squ90t003od000v900hrp.jpg\" alt=\"9b12780fj00squ90t003od000v900hrp\" width=\"1125\" height=\"639\" \/><\/p>\n<p>\u5728 Arena-Hard\u3001LiveBench\u3001LiveCodeBench \u548c GPQA-Diamond \u7b49\u57fa\u51c6\u6d4b\u8bd5\u4e2d\uff0cQwen2.5-Max \u7684\u8868\u73b0\u8d85\u8d8a\u4e86 DeepSeek V3\uff0c\u540c\u65f6\u5728 MMLU-Pro \u7b49\u5176\u4ed6\u8bc4\u4f30\u4e2d\u4e5f\u5c55\u73b0\u51fa\u4e86\u6781\u5177\u7ade\u4e89\u529b\u7684\u6210\u7ee9\u3002<\/p>\n<p>\u5728\u57fa\u5ea7\u6a21\u578b\u7684\u5bf9\u6bd4\u4e2d\uff0c\u7531\u4e8e\u65e0\u6cd5\u8bbf\u95ee GPT-4o \u548c Claude-3.5-Sonnet \u7b49\u95ed\u6e90\u6a21\u578b\u7684\u57fa\u5ea7\u6a21\u578b\uff0c\u963f\u91cc\u4e91\u5c06 Qwen2.5-Max \u4e0e\u76ee\u524d\u9886\u5148\u7684\u5f00\u6e90 MoE \u6a21\u578b DeepSeek V3\u3001\u6700\u5927\u7684\u5f00\u6e90\u7a20\u5bc6\u6a21\u578b Llama-3.1-405B\uff0c\u4ee5\u53ca\u540c\u6837\u4f4d\u5217\u5f00\u6e90\u7a20\u5bc6\u6a21\u578b\u524d\u5217\u7684 Qwen2.5-72B \u8fdb\u884c\u4e86\u5bf9\u6bd4\u3002\u5bf9\u6bd4\u7ed3\u679c\u5982\u4e0b\u56fe\u6240\u793a\uff1a<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-27925\" title=\"984bbaf7j00squ901003td000v900hkp\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/01\/984bbaf7j00squ901003td000v900hkp.jpg\" alt=\"984bbaf7j00squ901003td000v900hkp\" width=\"1125\" height=\"632\" \/><\/p>\n<p>\u6211\u4eec\u7684\u57fa\u5ea7\u6a21\u578b\u5728\u5927\u591a\u6570\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u90fd\u5c55\u73b0\u51fa\u4e86\u663e\u8457\u7684\u4f18\u52bf\u3002\u6211\u4eec\u76f8\u4fe1\uff0c\u968f\u7740\u540e\u8bad\u7ec3\u6280\u672f\u7684\u4e0d\u65ad\u8fdb\u6b65\uff0c\u4e0b\u4e00\u4e2a\u7248\u672c\u7684 Qwen2.5-Max \u5c06\u4f1a\u8fbe\u5230\u66f4\u9ad8\u7684\u6c34\u5e73\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>January 29, 2011 - On the occasion of the New Year, AliCloud announced its new Qwen 2.5-Max hyperscale MoE model, which can be accessed by way of APIs, or you can log in to Qwen Chat to experience it, for example, by talking to the model directly, or using artifacts, search and other functions. According to the introduction, Tongyi Qianqian Qwen 2.5-Max is trained using pre-training data of more than 20 trillion tokens and a well-designed post-training program. Performance Aliyun directly compared the performance of the command model (note: the command model is the model we usually use to have direct conversations). Comparison objects include DeepSeek V3, GPT-4o<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[146],"tags":[5653,331,334],"collection":[],"class_list":["post-27923","post","type-post","status-publish","format-standard","hentry","category-news","tag-moe","tag-331","tag-334"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/27923","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=27923"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/27923\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=27923"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=27923"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=27923"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=27923"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}