5万亿至10万亿参数模型，阿里真正要解决的不是“做大”而是“跑起来”

同花顺09-28 16:43

阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴（BABA）CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成更复杂、更长程的任务，向人工超智能迈进。先要厘清一个容易被忽略的事实：这是一项后续版本的规划目标，不是已经完成、也不是已经确定交付的模型。参数量是模型规模的一个指标，不能直接等同于智能水平。模型最终...

网页链接

免责声明：本文观点仅代表作者个人观点，不构成本平台的投资建议，本平台不对文章信息准确性、完整性和及时性做出任何保证，亦不对因使用或信赖文章信息引发的任何损失承担责任。

精彩评论

AI征战全球
09-28 20:09
AI征战全球
不懂请不要随便为了流量去请假

发表看法

{"i18n":{"language":"zh_CN"},"isChannel":false,"data":{"share":"https://www.laohunote.com/m/news/2670238505?lang=zh_CN&edition=full","thumbnail":"","is_english":false,"pubTime":"2026-09-28 16:43","share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","id":"2670238505","market":"us","top_or_hot":-1,"title":"5万亿至10万亿参数模型，阿里真正要解决的不是“做大”而是“跑起来”","media":"同花顺","content":"<div>\n<p>阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴（BABA）CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成更复杂、更长程的任务，向人工超智能迈进。先要厘清一个容易被忽略的事实：这是一项后续版本的规划目标，不是已经完成、也不是已经确定交付的模型。参数量是模型规模的一个指标，不能直接等同于智能水平。模型最终...</p>\n\n<a href=\"http://news.10jqka.com.cn/20260928/c680321169.shtml\">网页链接</a>\n\n</div>\n","source":"tonghuashun_highlight","html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>5万亿至10万亿参数模型，阿里真正要解决的不是“做大”而是“跑起来”</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\n5万亿至10万亿参数模型，阿里真正要解决的不是“做大”而是“跑起来”\n</h2>\n\n<h4 class=\"meta\">\n\n\n2026-09-28 16:43 北京时间&nbsp;&nbsp;&nbsp;<a href=http://news.10jqka.com.cn/20260928/c680321169.shtml><strong>同花顺</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴（BABA）CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成更复杂、更长程的任务，向人工超智能迈进。先要厘清一个容易被忽略的事实：这是一项后续版本的规划目标，不是已经完成、也不是已经确定交付的模型。参数量是模型规模的一个指标，不能直接等同于智能水平。模型最终...</p>\n\n<a href=\"http://news.10jqka.com.cn/20260928/c680321169.shtml\">网页链接</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","isBrief":false,"type":0,"news_type":1,"symbol":"LU1769817096.USD","symbol_name":"UBS (LUX) EQUITY SICAV - GLOBAL EMERG MARKETS OPPO \"P\" (USD) INC","start_time":0,"source_url":"http://news.10jqka.com.cn/20260928/c680321169.shtml","article_id":"2670238505","we_media_id":null,"thumbnails":[],"rights":{"source":"tonghuashun_highlight","url":"http://news.10jqka.com.cn/20260928/c680321169.shtml","rn_cache_url":null,"directOrigin":true},"url":"https://stock-news.laohu8.com/highlight/detail?id=2670238505","pubTimestamp":1790584980,"columns":[],"sourceInfo":{"source_id":"tonghuashun_highlight","name":"同花顺"},"weMediaInfo":null,"summary":"阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成 ...","collect":0,"end_time":0,"defaultTopTitle":"10jqka.com.cn","property":[],"viewcount":null,"language":"zh","relate_stocks":{"LU1769817096.USD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERG MARKETS OPPO \"P\" (USD) INC","BK4535":"淡马锡持仓","IE00BFMHRM44.USD":"NEUBERGER BERMAN GLOBAL EQUITY MEGATRENDS \"A\" (USD) ACC","LU0348816934.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT\" (USD)","LU1769817179.HKD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERG MARKETS OPPO \"P\" (HKD) INC","LU1282648689.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AMG\" (USD) INC","BK4538":"云计算","LU0918141705.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AM\" (HKD) INC","LU1823568750.SGD":"Fidelity Global Technology A-ACC SGD","BK4548":"巴美列捷福持仓","LU0047713382.USD":"BGF EMERGING MARKETS \"A2\" ACC","LU1051769294.HKD":"BGF EMERGING MARKETS EQUITY INCOME \"A6\"(HKDHDG) INC","LU1152091754.HKD":"UBS (LUX) EQUITY FUND - CHINA OPPORTUNITY (USD) \"PM\" (HKD) INC","LU2226246903.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT4\" (HKD) ACC","LU1048588211.SGD":"Blackrock Asian Dragon A2 SGD-H","BK4554":"元宇宙及AR概念","LU0072462343.USD":"贝莱德亚洲巨龙基金","LU0501845795.SGD":"瑞银大中华区股票基金P Acc SGD","LU0072913022.USD":"UBS (LUX) EQUITY FUND -  GREATER CHINA \"P\" (USD) ACC","BK4614":"Manus概念股","BK4585":"ETF&股票定投概念","LU1105468828.SGD":"Allianz Total Return Asian Equity AM DIS H2-SGD","KBAB":"2倍做多BABA ETF-KraneShares","BK4587":"ChatGPT概念","BK4558":"双十一","LU0106959298.USD":"UBS (LUX) EQUITY FUND - EMERGING MARKETS SUSTAINABLE LEADERS (USD) \"P\" (USD) ACC","LU0821914370.USD":"贝莱德亚洲成长领袖A2","LU1242518857.USD":"FULLERTON LUX FUNDS - ASIA ABSOLUTE ALPHA \"I\" (USD) ACC","LU0651947912.USD":"BGF EMERGING MARKETS EQUITY INCOME \"A5G\" (USD) INC","BK4588":"碎股","BK4565":"NFT概念","BK4607":"DeepSeek概念股","LU0328353924.USD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERGING MARKETS OPPORTUNITY (USD) \"P\" (USD) ACC","LU1880383366.USD":"东方汇理中国股票基金 A2 (C)","LU1046421795.USD":"富达环球科技A-ACC","LU0949170772.SGD":"Blackrock Global Equity Income A6 SGD-H","LU0251143458.SGD":"Fidelity Emerging Markets A-SGD","IE00BF5LJ272.USD":"Legg Mason Martin Currie - Global Emerging Markets A Acc USD","LU1003077747.HKD":"BGF GLOBAL EQUITY INCOME \"A6\" (HKDHDG) INC","LU0553294199.USD":"BGF GLOBAL EQUITY INCOME \"A5G\" (USD) INC","LU0880133367.SGD":"UBS (LUX) EQUITY FUND CHINA OPPORTUNITY USD \"P\" (SGD) ACC","BK4503":"景林资产持仓","09988":"阿里巴巴-W","LU0738911758.USD":"Blackrock Global Equity Income A6 USD","BK4502":"阿里概念","LU0918141887.USD":"安联亚洲实际收益股票基金","LU0348814723.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"A\" (USD) INC  NC","BK4505":"高瓴资本持仓","BK4504":"桥水持仓","LU0310800965.SGD":"FTIF - Templeton Global Balanced A Acc SGD","LU1242518931.SGD":"Fullerton Lux Funds - Asia Absolute Alpha A Acc SGD","LU0651946864.USD":"贝莱德新兴市场股票收益A2","LU1051768304.USD":"贝莱德新兴市场股票收益A6","BK4220":"综合零售","BK4524":"宅经济概念","LU0979878070.USD":"FULLERTON LUX FUNDS - ASIA ABSOLUTE ALPHA \"A\" (USD) ACC","BK4527":"明星科技股","LU1515016050.SGD":"Blackrock Emerging Markets Equity Income A6 SGD-H","BK4526":"热门中概股","LU1008478684.HKD":"UBS (LUX) EQUITY FUND -  CHINA OPPORTUNITY \"P\" (HKD) ACC","LU0456842615.SGD":"JPMorgan Funds - Emerging Markets Equity A (acc) SGD","LU0661504455.SGD":"Blackrock Global Equity Income A5 SGD-H","BK4122":"互联网与直销零售","BABA":"阿里巴巴","BK4575":"芯片概念","LU1048596156.SGD":"Blackrock Asian Growth Leaders A2 SGD-H","LU0545039389.USD":"BGF GLOBAL EQUITY INCOME \"A2\" ACC","LU1267930227.SGD":"TEMPLETON GLOBAL BALANCED \"AS\" (SGD) ACC A","BBYY":"GraniteShares YieldBOOST BABA ETF","LU1152091168.USD":"UBS (LUX) EQUITY FUND - CHINA OPPORTUNITY (USD) \"PM\" INC","LU1046422090.SGD":"Fidelity Pacific A-SGD","BABC":"Corgi BABA 2x Daily ETF","BK4579":"人工智能","LU0128525689.USD":"TEMPLETON GLOBAL BALANCED \"A\"(USD) ACC","ONLN":"ProShares Online Retail ETF","BABO":"BABA期权收益策略ETF-YieldMax","LU0067412154.USD":"UBS (LUX) EQUITY FUND -  CHINA OPPORTUNITY \"P\" (USD) ACC","LU0052756011.USD":"TEMPLETON GLOBAL BALANCED \"A\" (USD) INC","BABU":"Direxion Daily BABA Bull 2X ETF","DRAG":"中国龙ETF-Roundhill","BABW":"Roundhill BABA WeeklyPay ETF","BABX":"2倍做多BABA ETF-GraniteShares","CLIX":"ProShares Long Online/Short Stores ETF","BK4581":"高盛持仓","BK4543":"AI","LU1688375341.USD":"贝莱德中国灵活股票基金","LU0797268264.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT\" (HKD) ACC","LU2148611432.USD":"AZ ALLOCATION BALANCED BRAVE \"AAZ\" (USDHDG) ACC","BK4531":"中概回港概念","BK4534":"瑞士信贷持仓","BK4533":"AQR资本管理(全球第二大对冲基金)","IE00B0JY6N72.USD":"PINEBRIDGE GLOBAL EMERGING MARKETS FOCUS EQUITY \"A\" (USD) ACC","LU1235294995.USD":"FIDELITY GLOBAL TECHNOLOGY \"A\" (USDHDG) ACC"},"translate_title":"With a 5 trillion to 10 trillion parameter model, Alibaba's real goal is not to \"scale up\" but to \"get it running.\"","themeId":null,"isJumpTheme":false,"ttsUrl":"https://static.tigerbbs.com/3d88d63c7c7ed2348790382d2dbc646b","symbols_score_info":{"09988":1.97,"BBYY":0.6,"BABX":0.6,"BABC":0.6,"BABA":1.97,"KBAB":0.6,"DRAG":0.6,"BABU":0.6,"ONLN":0.6,"CLIX":0.6,"BABW":0.6,"BABO":0.6},"content_text":"阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴（BABA）CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成更复杂、更长程的任务，向人工超智能迈进。先要厘清一个容易被忽略的事实：这是一项后续版本的规划目标，不是已经完成、也不是已经确定交付的模型。参数量是模型规模的一个指标，不能直接等同于智能水平。模型最终表现还取决于训练数据质量、训练方法、架构设计、推理策略以及在真实任务中的稳定性。但5万亿至10万亿参数仍然是一个足以改变工程边界的数字。参数规模扩大，首先会把压力传导至显存、存储、网络互联和集群调度。训练阶段需要处理更大规模的参数、梯度和中间状态；推理阶段则要面对模型加载、并发请求和响应时延。模型能不能做出来是一回事，能不能在可接受的成本和时间内训练完成、部署上线并持续服务，是另一回事。可能的收益也不能简单概括为“参数越多，能力越强”。更大的模型有机会承载更复杂的知识关系和更长程的任务规划，但收益取决于新增参数是否被有效训练。若数据、算法和系统效率没有同步改善，规模扩张可能只是成本曲线的上移。阿里正在用芯片和云基础设施，为这一目标预留路线。平头哥发布的真武V900，被披露为上一代真武M890算力的3倍，配备216GB显存，片间互联带宽达到1200GB/s，单集群可扩展至50万卡，计划于2027年第一季度量产。这里的关键词不是单颗芯片的纸面参数，而是芯片、互联、软件和集群能否协同工作。M890已经被阿里披露用于支撑超过2万亿参数模型的高效推理。这个表述对应的是推理支持能力，不应被解读为已经完成了同等规模模型的训练。训练与推理对系统的要求不同：前者更强调大规模并行、通信效率和长时间稳定运行，后者更关注显存占用、吞吐、时延与单位Token成本。V900的意义也不只是替代某一款外部芯片。阿里披露的平头哥布局还包括倚天系列CPU、磐脉系列智能网卡和ICN互联芯片。芯片负责计算，网卡和互联决定数据如何流动，云平台则负责把这些资源调度给训练和推理任务。只看芯片算力，容易忽略通信瓶颈、故障率、软件适配和资源利用率对实际产出的影响。阿里为AI基础设施设定的长期刻度同样激进：到2032年，阿里云运营的全球数据中心规模超过20GW，较2022年增长10倍。这个目标显示，阿里押注的不是一次模型发布，而是模型、芯片和云的长期绑定。但数据中心规模扩大并不自动等于有效算力增加，电力供应、机房建设、存储网络和客户需求都需要同步兑现。更隐蔽的瓶颈在软件层。云栖大会披露的案例显示，Qwen3.8-Max曾自主调用EDA工具，在60多个小时内完成从编写电路描述、验证到后端物理实现的流程；相关模型还被用于适配和优化推理框架。这些案例说明，模型有望成为芯片设计和软件工程的工具，帮助降低部分研发环节的试错成本。但阶段性技术展示不能推导出超大模型已经具备自主研发能力。公开信息尚不足以确认上述实验的具体条件、人工介入程度、结果质量和可重复性，更不能据此证明5万亿至10万亿参数模型必然能够自我改进。真正的挑战，是把一次演示变成稳定、可审计、可复用的研发闭环。这也是阿里计划对AI产业生态最可能产生的影响：竞争将从单纯比拼模型榜单，转向比拼系统效率和供给能力。芯片厂商需要关注显存、互联和软件栈；云厂商需要提供更稳定的训练资源和推理服务；存储、网络、数据中心以及能源（850101）配套都可能受益于算力扩张。国家统计局公布的1—8月份数据已经显示，电子行业利润同比增长1.1倍，对规上工业企业利润增长贡献率达62%；光电子器件制造、半导体（881121）分立器件（884090）制造利润分别增长72.0%和51.8%。这类产业数据能够说明需求正在传导，但不能直接证明某一家公司的超大模型计划已经商业化。企业客户最终也不会为参数数量本身买单。阿里云副总裁霍嘉曾将企业AI项目失败归因于找错场景、高估数据质量，以及沿用传统信息化项目的做法。大模型只有在长程任务完成率、调用成本、响应速度和业务结果上形成优势，才可能转化为持续Token需求和云收入。参数规模与商业价值之间，还隔着一整套工程验证。判断这项计划能否落地，后续至少要看四组信号：第一，训练是否真正启动，模型架构、数据策略和训练阶段是否披露；第二，训练集群规模、V900量产进度以及云端供给能力能否兑现；第三，模型在长程任务、推理成本和真实客户中的表现，是否超越更小模型的性价比；第四，是否形成稳定的开发者、企业客户和工具链生态。所以，5万亿至10万亿参数更像一张工程路线图，而不是能力成绩单。阿里已经给出了模型、芯片和云的组合方案，也披露了M890、V900及20GW数据中心等支撑目标。接下来市场真正要验证的，不是数字还能不能继续变大，而是这套系统能否把参数转化为可持续、可负担、可复用的智能服务。","kind":"highlight","is_publish_news":true,"is_publish_highlight":false,"is_publish_live":false,"is_publish_wemedia":false,"editions":null,"column":"","sentiment":"0","news_tag":"","news_rank":0,"isVideo":false,"video":null,"symbols":[],"gpt_button":1,"need_auth":false,"need_login_tip":false,"code":"91000000","status":"200"},"commentList":[{"id":612671824572800,"gmtCreate":1790597345251,"gmtModify":1790597347418,"author":{"id":"4145482403543770","authorId":"4145482403543770","name":"AI征战全球","avatar":"https://static.tigerbbs.com/000a62cff47cb616d695cb48353c6db5","vip":1,"crmLevel":1,"crmLevelSwitch":0,"wearingBadges":[],"authorIdStr":"4145482403543770","idStr":"4145482403543770"},"htmlText":"不懂请不要随便为了流量去请假","listText":"不懂请不要随便为了流量去请假","text":"不懂请不要随便为了流量去请假","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/612671824572800","repostId":2670238505,"repostType":2,"repost":{"id":"2670238505","kind":"highlight","pubTimestamp":1790584980,"share":"https://www.laohunote.com/m/news/2670238505?lang=zh_CN&edition=full","pubTime":"2026-09-28 16:43","market":"us","language":"zh","title":"5万亿至10万亿参数模型，阿里真正要解决的不是“做大”而是“跑起来”","url":"https://stock-news.laohu8.com/highlight/detail?id=2670238505","media":"同花顺","summary":"阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成 ...","content":"<div>\n<p>阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴（BABA）CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成更复杂、更长程的任务，向人工超智能迈进。先要厘清一个容易被忽略的事实：这是一项后续版本的规划目标，不是已经完成、也不是已经确定交付的模型。参数量是模型规模的一个指标，不能直接等同于智能水平。模型最终...</p>\n\n<a href=\"http://news.10jqka.com.cn/20260928/c680321169.shtml\">网页链接</a>\n\n</div>\n","source":"tonghuashun_highlight","collect":0,"html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>5万亿至10万亿参数模型，阿里真正要解决的不是“做大”而是“跑起来”</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\n5万亿至10万亿参数模型，阿里真正要解决的不是“做大”而是“跑起来”\n</h2>\n\n<h4 class=\"meta\">\n\n\n2026-09-28 16:43 北京时间&nbsp;&nbsp;&nbsp;<a href=http://news.10jqka.com.cn/20260928/c680321169.shtml><strong>同花顺</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴（BABA）CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成更复杂、更长程的任务，向人工超智能迈进。先要厘清一个容易被忽略的事实：这是一项后续版本的规划目标，不是已经完成、也不是已经确定交付的模型。参数量是模型规模的一个指标，不能直接等同于智能水平。模型最终...</p>\n\n<a href=\"http://news.10jqka.com.cn/20260928/c680321169.shtml\">网页链接</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"","relate_stocks":{"LU1769817096.USD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERG MARKETS OPPO \"P\" (USD) INC","BK4535":"淡马锡持仓","IE00BFMHRM44.USD":"NEUBERGER BERMAN GLOBAL EQUITY MEGATRENDS \"A\" (USD) ACC","LU0348816934.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT\" (USD)","LU1769817179.HKD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERG MARKETS OPPO \"P\" (HKD) INC","LU1282648689.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AMG\" (USD) INC","BK4538":"云计算","LU0918141705.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AM\" (HKD) INC","LU1823568750.SGD":"Fidelity Global Technology A-ACC SGD","BK4548":"巴美列捷福持仓","LU0047713382.USD":"BGF EMERGING MARKETS \"A2\" ACC","LU1051769294.HKD":"BGF EMERGING MARKETS EQUITY INCOME \"A6\"(HKDHDG) INC","LU1152091754.HKD":"UBS (LUX) EQUITY FUND - CHINA OPPORTUNITY (USD) \"PM\" (HKD) INC","LU2226246903.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT4\" (HKD) ACC","LU1048588211.SGD":"Blackrock Asian Dragon A2 SGD-H","BK4554":"元宇宙及AR概念","LU0072462343.USD":"贝莱德亚洲巨龙基金","LU0501845795.SGD":"瑞银大中华区股票基金P Acc SGD","LU0072913022.USD":"UBS (LUX) EQUITY FUND -  GREATER CHINA \"P\" (USD) ACC","BK4614":"Manus概念股","BK4585":"ETF&股票定投概念","LU1105468828.SGD":"Allianz Total Return Asian Equity AM DIS H2-SGD","KBAB":"2倍做多BABA ETF-KraneShares","BK4587":"ChatGPT概念","BK4558":"双十一","LU0106959298.USD":"UBS (LUX) EQUITY FUND - EMERGING MARKETS SUSTAINABLE LEADERS (USD) \"P\" (USD) ACC","LU0821914370.USD":"贝莱德亚洲成长领袖A2","LU1242518857.USD":"FULLERTON LUX FUNDS - ASIA ABSOLUTE ALPHA \"I\" (USD) ACC","LU0651947912.USD":"BGF EMERGING MARKETS EQUITY INCOME \"A5G\" (USD) INC","BK4588":"碎股","BK4565":"NFT概念","BK4607":"DeepSeek概念股","LU0328353924.USD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERGING MARKETS OPPORTUNITY (USD) \"P\" (USD) ACC","LU1880383366.USD":"东方汇理中国股票基金 A2 (C)","LU1046421795.USD":"富达环球科技A-ACC","LU0949170772.SGD":"Blackrock Global Equity Income A6 SGD-H","LU0251143458.SGD":"Fidelity Emerging Markets A-SGD","IE00BF5LJ272.USD":"Legg Mason Martin Currie - Global Emerging Markets A Acc USD","LU1003077747.HKD":"BGF GLOBAL EQUITY INCOME \"A6\" (HKDHDG) INC","LU0553294199.USD":"BGF GLOBAL EQUITY INCOME \"A5G\" (USD) INC","LU0880133367.SGD":"UBS (LUX) EQUITY FUND CHINA OPPORTUNITY USD \"P\" (SGD) ACC","BK4503":"景林资产持仓","09988":"阿里巴巴-W","LU0738911758.USD":"Blackrock Global Equity Income A6 USD","BK4502":"阿里概念","LU0918141887.USD":"安联亚洲实际收益股票基金","LU0348814723.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"A\" (USD) INC  NC","BK4505":"高瓴资本持仓","BK4504":"桥水持仓","LU0310800965.SGD":"FTIF - Templeton Global Balanced A Acc SGD","LU1242518931.SGD":"Fullerton Lux Funds - Asia Absolute Alpha A Acc SGD","LU0651946864.USD":"贝莱德新兴市场股票收益A2","LU1051768304.USD":"贝莱德新兴市场股票收益A6","BK4220":"综合零售","BK4524":"宅经济概念","LU0979878070.USD":"FULLERTON LUX FUNDS - ASIA ABSOLUTE ALPHA \"A\" (USD) ACC","BK4527":"明星科技股","LU1515016050.SGD":"Blackrock Emerging Markets Equity Income A6 SGD-H","BK4526":"热门中概股","LU1008478684.HKD":"UBS (LUX) EQUITY FUND -  CHINA OPPORTUNITY \"P\" (HKD) ACC","LU0456842615.SGD":"JPMorgan Funds - Emerging Markets Equity A (acc) SGD","LU0661504455.SGD":"Blackrock Global Equity Income A5 SGD-H","BK4122":"互联网与直销零售","BABA":"阿里巴巴","BK4575":"芯片概念","LU1048596156.SGD":"Blackrock Asian Growth Leaders A2 SGD-H","LU0545039389.USD":"BGF GLOBAL EQUITY INCOME \"A2\" ACC","LU1267930227.SGD":"TEMPLETON GLOBAL BALANCED \"AS\" (SGD) ACC A","BBYY":"GraniteShares YieldBOOST BABA ETF","LU1152091168.USD":"UBS (LUX) EQUITY FUND - CHINA OPPORTUNITY (USD) \"PM\" INC","LU1046422090.SGD":"Fidelity Pacific A-SGD","BABC":"Corgi BABA 2x Daily ETF","BK4579":"人工智能","LU0128525689.USD":"TEMPLETON GLOBAL BALANCED \"A\"(USD) ACC","ONLN":"ProShares Online Retail ETF","BABO":"BABA期权收益策略ETF-YieldMax","LU0067412154.USD":"UBS (LUX) EQUITY FUND -  CHINA OPPORTUNITY \"P\" (USD) ACC","LU0052756011.USD":"TEMPLETON GLOBAL BALANCED \"A\" (USD) INC","BABU":"Direxion Daily BABA Bull 2X ETF","DRAG":"中国龙ETF-Roundhill","BABW":"Roundhill BABA WeeklyPay ETF","BABX":"2倍做多BABA ETF-GraniteShares","CLIX":"ProShares Long Online/Short Stores ETF","BK4581":"高盛持仓","BK4543":"AI","LU1688375341.USD":"贝莱德中国灵活股票基金","LU0797268264.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT\" (HKD) ACC","LU2148611432.USD":"AZ ALLOCATION BALANCED BRAVE \"AAZ\" (USDHDG) ACC","BK4531":"中概回港概念","BK4534":"瑞士信贷持仓","BK4533":"AQR资本管理(全球第二大对冲基金)","IE00B0JY6N72.USD":"PINEBRIDGE GLOBAL EMERGING MARKETS FOCUS EQUITY \"A\" (USD) ACC","LU1235294995.USD":"FIDELITY GLOBAL TECHNOLOGY \"A\" (USDHDG) ACC"},"source_url":"http://news.10jqka.com.cn/20260928/c680321169.shtml","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"2670238505","content_text":"阿里把下一阶段的大模型目标，直接推到了5万亿至10万亿参数。在云栖大会上，阿里巴巴（BABA）CEO吴泳铭表示，Qwen团队正在推进模型架构和数据优化研究，计划训练5—10T参数规模的新模型，目标是完成更复杂、更长程的任务，向人工超智能迈进。先要厘清一个容易被忽略的事实：这是一项后续版本的规划目标，不是已经完成、也不是已经确定交付的模型。参数量是模型规模的一个指标，不能直接等同于智能水平。模型最终表现还取决于训练数据质量、训练方法、架构设计、推理策略以及在真实任务中的稳定性。但5万亿至10万亿参数仍然是一个足以改变工程边界的数字。参数规模扩大，首先会把压力传导至显存、存储、网络互联和集群调度。训练阶段需要处理更大规模的参数、梯度和中间状态；推理阶段则要面对模型加载、并发请求和响应时延。模型能不能做出来是一回事，能不能在可接受的成本和时间内训练完成、部署上线并持续服务，是另一回事。可能的收益也不能简单概括为“参数越多，能力越强”。更大的模型有机会承载更复杂的知识关系和更长程的任务规划，但收益取决于新增参数是否被有效训练。若数据、算法和系统效率没有同步改善，规模扩张可能只是成本曲线的上移。阿里正在用芯片和云基础设施，为这一目标预留路线。平头哥发布的真武V900，被披露为上一代真武M890算力的3倍，配备216GB显存，片间互联带宽达到1200GB/s，单集群可扩展至50万卡，计划于2027年第一季度量产。这里的关键词不是单颗芯片的纸面参数，而是芯片、互联、软件和集群能否协同工作。M890已经被阿里披露用于支撑超过2万亿参数模型的高效推理。这个表述对应的是推理支持能力，不应被解读为已经完成了同等规模模型的训练。训练与推理对系统的要求不同：前者更强调大规模并行、通信效率和长时间稳定运行，后者更关注显存占用、吞吐、时延与单位Token成本。V900的意义也不只是替代某一款外部芯片。阿里披露的平头哥布局还包括倚天系列CPU、磐脉系列智能网卡和ICN互联芯片。芯片负责计算，网卡和互联决定数据如何流动，云平台则负责把这些资源调度给训练和推理任务。只看芯片算力，容易忽略通信瓶颈、故障率、软件适配和资源利用率对实际产出的影响。阿里为AI基础设施设定的长期刻度同样激进：到2032年，阿里云运营的全球数据中心规模超过20GW，较2022年增长10倍。这个目标显示，阿里押注的不是一次模型发布，而是模型、芯片和云的长期绑定。但数据中心规模扩大并不自动等于有效算力增加，电力供应、机房建设、存储网络和客户需求都需要同步兑现。更隐蔽的瓶颈在软件层。云栖大会披露的案例显示，Qwen3.8-Max曾自主调用EDA工具，在60多个小时内完成从编写电路描述、验证到后端物理实现的流程；相关模型还被用于适配和优化推理框架。这些案例说明，模型有望成为芯片设计和软件工程的工具，帮助降低部分研发环节的试错成本。但阶段性技术展示不能推导出超大模型已经具备自主研发能力。公开信息尚不足以确认上述实验的具体条件、人工介入程度、结果质量和可重复性，更不能据此证明5万亿至10万亿参数模型必然能够自我改进。真正的挑战，是把一次演示变成稳定、可审计、可复用的研发闭环。这也是阿里计划对AI产业生态最可能产生的影响：竞争将从单纯比拼模型榜单，转向比拼系统效率和供给能力。芯片厂商需要关注显存、互联和软件栈；云厂商需要提供更稳定的训练资源和推理服务；存储、网络、数据中心以及能源（850101）配套都可能受益于算力扩张。国家统计局公布的1—8月份数据已经显示，电子行业利润同比增长1.1倍，对规上工业企业利润增长贡献率达62%；光电子器件制造、半导体（881121）分立器件（884090）制造利润分别增长72.0%和51.8%。这类产业数据能够说明需求正在传导，但不能直接证明某一家公司的超大模型计划已经商业化。企业客户最终也不会为参数数量本身买单。阿里云副总裁霍嘉曾将企业AI项目失败归因于找错场景、高估数据质量，以及沿用传统信息化项目的做法。大模型只有在长程任务完成率、调用成本、响应速度和业务结果上形成优势，才可能转化为持续Token需求和云收入。参数规模与商业价值之间，还隔着一整套工程验证。判断这项计划能否落地，后续至少要看四组信号：第一，训练是否真正启动，模型架构、数据策略和训练阶段是否披露；第二，训练集群规模、V900量产进度以及云端供给能力能否兑现；第三，模型在长程任务、推理成本和真实客户中的表现，是否超越更小模型的性价比；第四，是否形成稳定的开发者、企业客户和工具链生态。所以，5万亿至10万亿参数更像一张工程路线图，而不是能力成绩单。阿里已经给出了模型、芯片和云的组合方案，也披露了M890、V900及20GW数据中心等支撑目标。接下来市场真正要验证的，不是数字还能不能继续变大，而是这套系统能否把参数转化为可持续、可负担、可复用的智能服务。","news_type":1,"symbols_score_info":{"09988":1.97,"BBYY":0.6,"BABX":0.6,"BABC":0.6,"BABA":1.97,"KBAB":0.6,"DRAG":0.6,"BABU":0.6,"ONLN":0.6,"CLIX":0.6,"BABW":0.6,"BABO":0.6}},"isVote":1,"likeStatus":false,"favoriteStatus":false,"reportStatus":false,"tweetType":1,"langContent":"CN"}],"isCommentEnd":true,"newsSizeData":{"likeSize":0,"commentSize":1,"repostSize":1,"favoriteSize":0,"likeStatus":false,"favoriteStatus":false},"APP":{"userAgent":"Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)","isDev":false,"isTTM":false,"isLaohu8Offline":false,"tenantId":"TBCN","deviceId":"web-server-community-laohu8-v3","version":"4.50.1","shortVersion":"4.50.1","platform":"web","vendor":"web","appName":"laohu8","isIOS":false,"isAndroid":false,"isTiger":false,"isTHS":false,"isWeiXin":false,"isWeiXinMini":false,"isWeiBo":false,"isQQ":false,"isBaiduSwan":false,"isBaiduBox":false,"isDingTalk":false,"isToutiao":false,"isOnePlus":false,"isHuaWei":false,"isXiaomi":false,"isXiaomiWebView":false,"isOppo":false,"isVivo":false,"isSamsung":false,"isMobile":false},"href":"/m/news/2670238505","isCrawlerRequest":true}