{"id":242221,"date":"2026-11-16T07:50:10","date_gmt":"2026-11-16T07:50:10","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/242221\/"},"modified":"2026-11-16T07:50:10","modified_gmt":"2026-11-16T07:50:10","slug":"alibabas-giant-qwen4-max-model-trails-claude-gpt-and-moonshot-ai","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/242221\/","title":{"rendered":"Alibaba&#8217;s Giant Qwen4-Max Model Trails Claude, GPT and Moonshot AI"},"content":{"rendered":"<p class=\"tt-google-pref-wrap\"><a class=\"tt-google-pref\" href=\"https:\/\/www.google.com\/preferences\/source?q=trendingtopics.eu\" target=\"_blank\" rel=\"noopener nofollow\">Set Trending Topics as a preferred source on Google.<\/a><\/p>\n<p>Bigger does not automatically mean better: Alibaba has unveiled Qwen4-Max, its largest A.I. model yet, with 2.4 trillion parameters and a context window of one million tokens. The model is meant to work on projects around the clock as an \u201cA.I. teammate,\u201d from 3D models and games to legal reviews, the Qwen team wrote in its <a href=\"https:\/\/qwen.ai\/blog?id=qwen4-max\" rel=\"nofollow noopener\" target=\"_blank\">launch blog<\/a>. In independent rankings, however, Qwen4-Max lands well behind the top models from the United States and also behind its Chinese rival Moonshot.<\/p>\n<p>What Qwen4-Max Can Do<\/p>\n<p>Qwen4-Max is a so-called mixture-of-experts model: Of its 2.4 trillion parameters, only 71 billion are active per token, plus 328 billion parameters for a new N-gram embedding. Besides text, the model also understands images, and it is a reasoning model. According to Alibaba, it uses significantly fewer tokens for the same tasks than its predecessor, Qwen3.8-Max. On the company\u2019s own Work360 benchmark, the cost per task fell from $4.43 (about 3.98 euros) to $1.67 (about 1.50 euros).<\/p>\n<p>Alibaba presents a whole series of real-world examples:<\/p>\n<p>Industry: At two automotive suppliers, the model reportedly cut the programming effort for a single workpiece from about 475 minutes to 24.<br \/>\nLaw: For pre-I.P.O. legal due diligence, Qwen4-Max sorted 198 working papers and delivered an 86-page report.<br \/>\nResearch: In a 48-hour autonomous run, the model designed a new coolant for data centers, which an automated lab then synthesized.<\/p>\n<p>Qwen4-Max is available through QwenCloud via API and in Alibaba\u2019s QwenWork, Qoder and QwenStudio apps.<\/p>\n<p>Open Weights Only for the Smaller Models<\/p>\n<p>Anyone hoping to run Qwen4-Max on their own servers is out of luck: Artificial Analysis lists the model as proprietary, and its weights are not available for download. Alibaba is thus sticking with its practice of offering its Max models only through its own cloud, as it has since Qwen2.5-Max. The smaller models in the Qwen4 family, by contrast, are to be released as open source, built on the same architecture and trained with the same techniques, the Qwen team said.<\/p>\n<p>With its open models, Alibaba has built a large developer community and recently even pushed <a href=\"https:\/\/www.trendingtopics.eu\/tencent-hy4-open-weight-alibaba-qwen-en\/\" rel=\"nofollow noopener\" target=\"_blank\">Tencent to follow suit<\/a>. Jensen Huang recently underlined how important Chinese open-weight models have become: According to the former Nvidia chief, 80 percent of U.S. start-ups use <a href=\"https:\/\/www.trendingtopics.eu\/nvidia-without-huang-open-weights-circular-deals-china\/\" rel=\"nofollow noopener\" target=\"_blank\">Chinese open models<\/a>. The strongest open-weight model outside China is currently <a href=\"https:\/\/www.trendingtopics.eu\/inkling-2-thinking-machines-mistral-open-weight-en\/\" rel=\"nofollow noopener\" target=\"_blank\">Inkling 2<\/a> from Mira Murati\u2019s Thinking Machines.<\/p>\n<p>Ninth Among the Leaders<\/p>\n<p>Independent tests paint a more sober picture. On the Intelligence Index from <a href=\"https:\/\/artificialanalysis.ai\/models\/qwen4-max\" rel=\"nofollow noopener\" target=\"_blank\">Artificial Analysis<\/a>, Qwen4-Max scores 43 points. Among the 10 leading models the service compares directly, that is only enough for ninth place:<\/p>\n<p>Rank<br \/>\nModel<br \/>\nDeveloper<br \/>\nIntelligence Index<\/p>\n<p>1<br \/>\nClaude Fable 5.2<br \/>\nAnthropic<br \/>\n65<\/p>\n<p>2<br \/>\nGemini 4 Carbon<br \/>\nGoogle<br \/>\n60<\/p>\n<p>3<br \/>\nClaude Opus 5.5<br \/>\nAnthropic<br \/>\n59<\/p>\n<p>4<br \/>\nGPT-6 Astra<br \/>\nOpenAI<br \/>\n57<\/p>\n<p>5<br \/>\nGPT-6.1 Sol<br \/>\nOpenAI<br \/>\n55<\/p>\n<p>6<br \/>\nMuse Wonder Full<br \/>\nMeta<br \/>\n50<\/p>\n<p>7<br \/>\nKimi K3.1 Pro<br \/>\nMoonshot AI<br \/>\n46<\/p>\n<p>8<br \/>\nMuse Wonder<br \/>\nMeta<br \/>\n44<\/p>\n<p>9<br \/>\nQwen4-Max<br \/>\nAlibaba<br \/>\n43<\/p>\n<p>10<br \/>\nGrok 4.7<br \/>\nxAI<br \/>\n35<\/p>\n<p>In the intelligence ranking for its model class, Artificial Analysis places Qwen4-Max 38th out of 81, just above the median of 41 points. That leaves Moonshot, with Kimi K3.1 Pro, as the best-performing Chinese developer.<\/p>\n<p>Ahead in Law and Agents, Behind in Coding<\/p>\n<p>Even in Alibaba\u2019s own comparisons, Qwen4-Max does not lead everywhere. The model is strong at legal agent work and long tasks, while Anthropic, OpenAI and Moonshot are ahead in agentic coding. A selection of the figures Alibaba published:<\/p>\n<p>Benchmark<br \/>\nQwen4-Max<br \/>\nClaude Opus 5.5<br \/>\nGPT-6.1 Sol<br \/>\nGPT-6 Astra<br \/>\nKimi K3.1<\/p>\n<p>HLAB-Vals 1.1 (legal)<br \/>\n21.3<br \/>\n9.2<br \/>\n6.3<br \/>\n7.1<br \/>\n11.7<\/p>\n<p>DeepSWE v1.1 (coding)<br \/>\n75.7<br \/>\n74.2<br \/>\n75.2<br \/>\n74.1<br \/>\n75.5<\/p>\n<p>Terminal-Bench 4.0<br \/>\n48.5<br \/>\n59.6<br \/>\n56.1<br \/>\n59.1<br \/>\n67.2<\/p>\n<p>ProgramBench<br \/>\n31.5<br \/>\n65.0<br \/>\n52.5<br \/>\n55.0<br \/>\n25.0<\/p>\n<p>Humanity\u2019s Last Exam<br \/>\n51.7<br \/>\n59.8<br \/>\n54.0<br \/>\n53.6<br \/>\n51.1<\/p>\n<p>APEX-Agents v1.1<br \/>\n72.1<br \/>\n73.5<br \/>\n60.0<br \/>\n64.7<br \/>\n65.5<\/p>\n<p>Cheap but Chatty<\/p>\n<p>On price, Alibaba plays to its strengths. According to Artificial Analysis, Qwen4-Max costs $1.50 (about 1.35 euros) per million input tokens and $4.50 (about 4.04 euros) per million output tokens, with a 95 percent discount on cached inputs. A task from the Intelligence Index costs an average of $3.27, compared with $10.04 for Claude Opus 5.5; only GPT-6.1 Sol is cheaper, at $1.27. In return, Qwen4-Max talks a lot: It generated 127 million output tokens for the test, far more than the median of 78.6 million.<\/p>\n<p>Alibaba does not disclose what hardware it used to train the model. The question is sensitive since a forensic analysis found <a href=\"https:\/\/www.trendingtopics.eu\/chinese-ai-models-blackwell-fingerprint-nvidia\/\" rel=\"nofollow noopener\" target=\"_blank\">traces of banned Nvidia chips<\/a> in the newest models from DeepSeek, Moonshot, MiniMax and Alibaba\u2019s affiliate Ant Group.<\/p>\n<p>\t\t\t<a href=\"https:\/\/rankmystartup.com\/\" target=\"_blank\" rel=\"nofollow noopener\"><br \/>\n\t\t\t\t<img decoding=\"async\" src=\"https:\/\/www.trendingtopics.eu\/wp-content\/themes\/trendingtopics\/images\/rankmystartup.webp\" alt=\"Rank My Startup: Erobere die Liga der Top Founder!\" class=\"display-light\"\/><br \/>\n\t\t\t<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"Set Trending Topics as a preferred source on Google. Bigger does not automatically mean better: Alibaba has unveiled&hellip;\n","protected":false},"author":2,"featured_media":242222,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[53,3154,182],"class_list":["post-242221","post","type-post","status-publish","format-standard","has-post-thumbnail","category-anthropic","tag-anthropic","tag-anthropic-claude","tag-claude"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/242221","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=242221"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/242221\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/242222"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=242221"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=242221"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=242221"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}