林俊旸阿里离职后发长文：大模型演进路线正从“推理型”全面转向“智能体”

凤凰网科技03-27

凤凰网科技讯 3月26日，前阿里千问技术负责人林俊旸离职后发表长文，明确指出AI大模型的发展路线正在经历重大跨越，核心竞争焦点正从“推理型思考（Reasoning Thinking）”全面转向“智能体思考（Agentic Thinking）”。文章复盘了以OpenAI o1和DeepSeek-R1为代表的第一波推理模型浪潮，指出这标志着行业从扩大预训练规模，正式步入扩大强化学习（RL）后训练规模的...

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

我们需要你的真知灼见来填补这片空白

发表看法

{"i18n":{"language":"zh_CN"},"isChannel":false,"data":{"share":"https://www.laohu8.com/m/news/2622618892?lang=zh_CN&edition=full","thumbnail":"","is_english":false,"pubTime":"2026-03-27 06:33","share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","id":"2622618892","market":"us","top_or_hot":-1,"title":"林俊旸阿里离职后发长文：大模型演进路线正从“推理型”全面转向“智能体”","media":"凤凰网科技","content":"<div>\n<p>凤凰网科技讯 3月26日，前阿里千问技术负责人林俊旸离职后发表长文，明确指出AI大模型的发展路线正在经历重大跨越，核心竞争焦点正从“推理型思考（Reasoning Thinking）”全面转向“智能体思考（Agentic Thinking）”。文章复盘了以OpenAI o1和DeepSeek-R1为代表的第一波推理模型浪潮，指出这标志着行业从扩大预训练规模，正式步入扩大强化学习（RL）后训练规模的...</p>\n\n<a href=\"https://tech.ifeng.com/c/8rpCV4zWK5v\">网页链接</a>\n\n</div>\n","source":"ifeng_tech","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>林俊旸阿里离职后发长文：大模型演进路线正从“推理型”全面转向“智能体”</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\">\n林俊旸阿里离职后发长文：大模型演进路线正从“推理型”全面转向“智能体”\n</h2>\n\n<h4 class=\"meta\">\n\n\n2026-03-27 06:33 北京时间&nbsp;&nbsp;&nbsp;<a href=https://tech.ifeng.com/c/8rpCV4zWK5v><strong>凤凰网科技</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>凤凰网科技讯 3月26日，前阿里千问技术负责人林俊旸离职后发表长文，明确指出AI大模型的发展路线正在经历重大跨越，核心竞争焦点正从“推理型思考（Reasoning Thinking）”全面转向“智能体思考（Agentic Thinking）”。文章复盘了以OpenAI o1和DeepSeek-R1为代表的第一波推理模型浪潮，指出这标志着行业从扩大预训练规模，正式步入扩大强化学习（RL）后训练规模的...</p>\n\n<a href=\"https://tech.ifeng.com/c/8rpCV4zWK5v\">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":"LU0880133367.SGD","symbol_name":"UBS (LUX) EQUITY FUND CHINA OPPORTUNITY USD \"P\" (SGD) ACC","start_time":0,"source_url":"https://tech.ifeng.com/c/8rpCV4zWK5v","article_id":"2622618892","we_media_id":null,"thumbnails":[],"rights":null,"url":"https://stock-news.laohu8.com/highlight/detail?id=2622618892","pubTimestamp":1774564410,"columns":[],"sourceInfo":{"source_id":"ifeng_tech","name":"凤凰网财经"},"weMediaInfo":null,"summary":"凤凰网科技讯 3月26日，前阿里千问技术负责人林俊旸离职后发表长文，明确指出AI大模型的发展路线正在经历重大跨越，核心竞争焦点正从“推理型思考 ”全面转向“智能体思考”。林俊旸在文中深度剖析了行业内尝试“融合思考与指令模式”所面临的落地困境。针对下一阶段的技术演进，林俊旸断言，单纯延长模型内部推理轨迹的时代即将过去，未来的主导将是在与环境交互中持续迭代计划的智能体思考。","collect":0,"end_time":0,"defaultTopTitle":"ifeng.com","property":[],"viewcount":null,"language":"zh","relate_stocks":{"LU0880133367.SGD":"UBS (LUX) EQUITY FUND CHINA OPPORTUNITY USD \"P\" (SGD) ACC","LU0553294199.USD":"BGF GLOBAL EQUITY INCOME \"A5G\" (USD) INC","BK4524":"宅经济概念","BK4220":"综合零售","BK4527":"明星科技股","LU0918141887.USD":"安联亚洲实际收益股票基金","LU0738911758.USD":"Blackrock Global Equity Income A6 USD","LU0348814723.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"A\" (USD) INC  NC","BK4526":"热门中概股","BK4122":"互联网与直销零售","LU0251143458.SGD":"Fidelity Emerging Markets A-SGD","BK4575":"芯片概念","LU1046421795.USD":"富达环球科技A-ACC","BABO":"BABA期权收益策略ETF-YieldMax","BABU":"Direxion Daily BABA Bull 2X ETF","LU0310800965.SGD":"FTIF - Templeton Global Balanced A Acc SGD","LU0651946864.USD":"贝莱德新兴市场股票收益A2","LU1242518931.SGD":"Fullerton Lux Funds - Asia Absolute Alpha A Acc SGD","BK4614":"Manus概念股","BK4579":"人工智能","BABW":"Roundhill BABA WeeklyPay ETF","LU1880383366.USD":"东方汇理中国股票基金 A2 (C)","LU1051768304.USD":"贝莱德新兴市场股票收益A6","BABX":"2倍做多BABA ETF-GraniteShares","LU0979878070.USD":"FULLERTON LUX FUNDS - ASIA ABSOLUTE ALPHA \"A\" (USD) ACC","LU1008478684.HKD":"UBS (LUX) EQUITY FUND -  CHINA OPPORTUNITY \"P\" (HKD) ACC","LU1515016050.SGD":"Blackrock Emerging Markets Equity Income A6 SGD-H","LU0456842615.SGD":"JPMorgan Funds - Emerging Markets Equity A (acc) SGD","BK4581":"高盛持仓","LU0661504455.SGD":"Blackrock Global Equity Income A5 SGD-H","BK4543":"AI","BK4531":"中概回港概念","LU0545039389.USD":"BGF GLOBAL EQUITY INCOME \"A2\" ACC","LU1048596156.SGD":"Blackrock Asian Growth Leaders A2 SGD-H","BK4534":"瑞士信贷持仓","LU1267930227.SGD":"TEMPLETON GLOBAL BALANCED \"AS\" (SGD) ACC A","BK4533":"AQR资本管理(全球第二大对冲基金)","KBAB":"2倍做多BABA ETF-KraneShares","LU1152091168.USD":"UBS (LUX) EQUITY FUND - CHINA OPPORTUNITY (USD) \"PM\" INC","LU1046422090.SGD":"Fidelity Pacific A-SGD","LU0128525689.USD":"TEMPLETON GLOBAL BALANCED \"A\"(USD) ACC","BK4535":"淡马锡持仓","BBYY":"GraniteShares YieldBOOST BABA ETF","LU0501845795.SGD":"瑞银大中华区股票基金P Acc SGD","LU0067412154.USD":"UBS (LUX) EQUITY FUND -  CHINA OPPORTUNITY \"P\" (USD) ACC","LU0052756011.USD":"TEMPLETON GLOBAL BALANCED \"A\" (USD) INC","BK4538":"云计算","BK4548":"巴美列捷福持仓","LU1823568750.SGD":"Fidelity Global Technology A-ACC SGD","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","BK4554":"元宇宙及AR概念","IE00B0JY6N72.USD":"PINEBRIDGE GLOBAL EMERGING MARKETS FOCUS EQUITY \"A\" (USD) ACC","LU1235294995.USD":"FIDELITY GLOBAL TECHNOLOGY \"A\" (USDHDG) ACC","LU1769817096.USD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERG MARKETS OPPO \"P\" (USD) INC","BK4585":"ETF&股票定投概念","IE00BFMHRM44.USD":"NEUBERGER BERMAN GLOBAL EQUITY MEGATRENDS \"A\" (USD) ACC","LU1769817179.HKD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERG MARKETS OPPO \"P\" (HKD) INC","LU0348816934.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT\" (USD)","LU1282648689.USD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AMG\" (USD) INC","BK4587":"ChatGPT概念","BK4558":"双十一","LU0047713382.USD":"BGF EMERGING MARKETS \"A2\" ACC","LU1051769294.HKD":"BGF EMERGING MARKETS EQUITY INCOME \"A6\"(HKDHDG) INC","LU2226246903.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AT4\" (HKD) ACC","LU1152091754.HKD":"UBS (LUX) EQUITY FUND - CHINA OPPORTUNITY (USD) \"PM\" (HKD) INC","BK4588":"碎股","LU1048588211.SGD":"Blackrock Asian Dragon A2 SGD-H","LU0072462343.USD":"贝莱德亚洲巨龙基金","BK4565":"NFT概念","LU0072913022.USD":"UBS (LUX) EQUITY FUND -  GREATER CHINA \"P\" (USD) ACC","LU1105468828.SGD":"Allianz Total Return Asian Equity AM DIS H2-SGD","LU0918141705.HKD":"ALLIANZ TOTAL RETURN ASIAN EQUITY \"AM\" (HKD) INC","BK4607":"DeepSeek概念股","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","BK4503":"景林资产持仓","LU0651947912.USD":"BGF EMERGING MARKETS EQUITY INCOME \"A5G\" (USD) INC","BK4502":"阿里概念","BK4505":"高瓴资本持仓","IE00BF5LJ272.USD":"Legg Mason Martin Currie - Global Emerging Markets A Acc USD","BK4504":"桥水持仓","LU0328353924.USD":"UBS (LUX) EQUITY SICAV - GLOBAL EMERGING MARKETS OPPORTUNITY (USD) \"P\" (USD) ACC","LU0949170772.SGD":"Blackrock Global Equity Income A6 SGD-H","LU1003077747.HKD":"BGF GLOBAL EQUITY INCOME \"A6\" (HKDHDG) INC"},"translate_title":"Lin Junyang Ali posted a long article after his departure: The evolution route of large models is shifting from \"reasoning\" to \"intelligent body\"","themeId":null,"isJumpTheme":false,"ttsUrl":null,"symbols_score_info":{"ALBmain":0.6,"BBYY":0.6,"BABW":0.6,"BABX":0.6,"BABU":0.6,"BABO":0.6,"KBAB":0.6},"content_text":"凤凰网科技讯 3月26日，前阿里千问技术负责人林俊旸离职后发表长文，明确指出AI大模型的发展路线正在经历重大跨越，核心竞争焦点正从“推理型思考（Reasoning Thinking）”全面转向“智能体思考（Agentic Thinking）”。文章复盘了以OpenAI o1和DeepSeek-R1为代表的第一波推理模型浪潮，指出这标志着行业从扩大预训练规模，正式步入扩大强化学习（RL）后训练规模的新阶段，数学与代码等可验证领域成为优化模型正确性的核心试金石。林俊旸在文中深度剖析了行业内尝试“融合思考与指令模式”所面临的落地困境。他透露，千问团队曾试图通过Qwen3打造支持混合思考模式的系统，但在实际推进中发现，指令模型追求极简与低延迟，而思考模型需要消耗大量Token进行复杂推演，两者在数据分布和行为目标上存在根本冲突。若数据筛选不当，强行融合往往会导致模型在两端表现平庸。基于商业客户对高吞吐量和低成本的真实需求，Qwen在后续的2507版本中选择推出了分离的30B和235B指令与思考变体。与之形成对比的是，Anthropic和DeepSeek等厂商则继续在统合推理与工具调用的混合架构上进行探索。针对下一阶段的技术演进，林俊旸断言，单纯延长模型内部推理轨迹的时代即将过去，未来的主导将是在与环境交互中持续迭代计划的智能体思考。他指出，智能体强化学习（Agentic RL）彻底改变了原有的技术栈要求，训练与推理必须实现更纯粹的解耦。随着大模型获得搜索、代码执行等工具权限，防范奖励作弊（Reward Hacking）将成为极其危险的挑战。未来的行业护城河将不再局限于算法本身，而是转移至高质量环境设计、防作弊协议以及多智能体协同编排等系统工程能力上。","kind":"news","is_publish_news":true,"is_publish_highlight":false,"is_publish_live":false,"is_publish_wemedia":null,"editions":null,"column":"","sentiment":"0","news_tag":"","news_rank":0,"isVideo":false,"video":null,"symbols":[],"gpt_button":0,"need_auth":false,"need_login_tip":false,"code":"91000000","status":"200"},"commentList":[],"isCommentEnd":true,"newsSizeData":{"likeSize":0,"commentSize":0,"repostSize":0,"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,"tenantId":"TBCN","deviceId":"web-server-community-laohu8-v3","version":"4.42.0","shortVersion":"4.42.0","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/2622618892","isCrawlerRequest":true}