Workflow
BABA(BABA)
icon
Search documents
上新!阿里Qwen3.5发布,AI技术与应用双线进击
2月16日,阿里巴巴悄然亮出AI领域的"春节档杀手锏"——全新一代开源大模型Qwen3.5-Plus发布。通过底层模型架构的全面革新,Qwen3.5-Plus版本总参 数为3970亿,激活仅170亿,以小胜大,性能超过万亿参数的Qwen3-Max模型,部署显存占用降低60%,推理效率大幅提升,最大推理吞吐量可提升至19 倍。与此同时,成本下降,Qwen3.5-Plus的API价格每百万Token低至0.8元,仅为Gemini 3 Pro的1/18。 图片来源:"千问大模型"微信公众号 值得一提的是,在底层技术突破的同时,截至目前,阿里AI应用战绩同样斐然。 2月17日,据阿里巴巴介绍,"千问请客"活动期间,有超过1.3亿人在千问点奶茶、囤年货、买电影票、买门票、订机票酒店。AI技术与应用双线进击背 后,是阿里从技术自主可控到生态协同发力的全链布局,彰显中国科技巨头在全球AI竞赛中的硬核实力。 阿里AI双线进击 2026年的AI"春节档"火热异常,阿里Qwen3.5凭借技术和性价比,将AI竞赛推向新高度。作为阿里全新一代基模,Qwen3.5-Plus一经开源,便成为全球最 强开源模型之一,性能直接媲美闭源第一梯 ...
技术为骨、生态为翼:千问“春节战役”引发的AI生活变革
Xin Hua Ri Bao· 2026-02-17 04:01
2026年的春节,中国互联网的底色是人工智能。"春节AI大战"已上演半个月,阿里巴巴发起了一场堪称教科书的AI"闪电战",完成了从基础大模型到AI应 用的双线进击。 2月17日大年初一凌晨,千问发布数据显示,春节活动期间,全国超过1.3亿人第一次体验AI购物,说了50亿次"千问帮我",千问一跃成为国民级AI助手。 可以说,阿里AI在这个春节获得了两条战线的全面胜利:千问跻身成为"国民AI",Qwen3.5则登顶全球最强开源模型。这场胜利,是阿里十年磨一剑的战 略兑现。"通云哥"组成的"黄金三角"协同作战,构筑了全球罕见的"自研芯片+云平台+大模型"全栈垂直整合体系。拥有自研AI芯片"真武",意味着阿里可 为千问量身定制硬件架构;而阿里云则如同"最懂模型的调度员",实现资源效率最大化。这种软硬一体的"全栈AI"模式,正是谷歌TPU与Gemini成功逻辑 的翻版,也是阿里掌握核心算力定义权的关键一步。 放眼全球,唯有阿里同时具备顶尖模型能力与庞大实体生态,真正补齐了"从意图到执行"的最后一块拼图。这场由千问点燃的AI生活化浪潮,或将重塑整 个行业的竞争格局——技术为骨,生态为翼,谁能在真实世界中让AI"落地生根" ...
Alibaba Group unveils Qwen3.5 as China's chatbot race shifts to AI agents
CNBC· 2026-02-17 03:11
Core Insights - Alibaba has launched its latest AI model series, Qwen3, which integrates traditional large language model capabilities with advanced dynamic reasoning to enhance its competitive edge in the Chinese AI market [1] Model Features - The Qwen3.5 AI model is available in an open-weight version for users to download, run, fine-tune, and deploy on their own infrastructure, alongside a hosted version on Alibaba's servers [2] - Qwen3.5 features improvements in performance and cost, with native multimodal capabilities that allow it to process text, images, and video simultaneously [3] - The model supports new coding and agentic capabilities, compatible with popular open-source AI agents like those from OpenClaw [3] Competitive Landscape - Local competitors such as ByteDance and Zhipu AI have also released upgraded models aimed at enhancing agent capabilities [5] - Qwen3.5 has 397 billion parameters, which is fewer than its predecessor, but it has shown significant performance improvements based on self-reported benchmarks [5] - Benchmark tests indicate that Qwen3.5's performance is comparable to leading models from OpenAI, Anthropic, and Google DeepMind, although these comparisons are self-reported [6] Language Support and Future Developments - The new Qwen3.5 models support 201 languages and dialects, an increase from the previous generation's 82 [7] - Alibaba is expected to release more open-weight models during the Chinese New Year, as indicated by the technical lead of Alibaba Cloud's Qwen team [7] Industry Trends - The release of Anthropic's Claude AI agent tools has prompted other American AI companies to accelerate the development of agentic capabilities [8] - Industry leaders, including OpenAI and Google DeepMind, acknowledge that Chinese AI models are rapidly advancing and are only months behind their Western counterparts [8]
Seedance 2.0画出骏马与繁花,人形机器人首次开口演小品,豆包、千问、元宝争相用红包“炸开”用户手机揭秘2026年春晚AI“烟火气”
Mei Ri Jing Ji Xin Wen· 2026-02-17 02:17
Core Insights - The 2026 Spring Festival Gala showcased the evolution of AI technology, with ByteDance's Seedance 2.0 model creating stunning visual effects that captivated millions of viewers [1][2][4] - AI has transitioned from a supporting role to a central figure in content creation, stage presentation, and live broadcasting, marking a significant shift in how technology integrates with traditional cultural events [3][5] Group 1: AI in Content Creation - The Seedance 2.0 model was used for visual content generation in several performances, including "The Song of the Wind" and "The Flower God," demonstrating its capability in producing complex imagery [4] - AI models have been deeply integrated into the creative process of the Spring Festival Gala, with various AI technologies enhancing the overall production quality [3][4] Group 2: AI in Performance and Interaction - A humanoid robot equipped with a voice model made its debut in a language-based skit, marking a historic moment in the integration of AI into live performances [5][9] - The use of real-time subtitle technology during the live broadcast improved accessibility for hearing-impaired viewers, showcasing AI's role in enhancing user experience [9][10] Group 3: AI Applications and User Engagement - Major tech companies engaged in a "red envelope war" during the Spring Festival, using AI applications to attract users and promote engagement through cash giveaways [11][12][19] - The launch of various promotional activities, such as the "30 billion yuan free order" campaign by Alibaba's Qianwen app, significantly exceeded expectations in user participation [16][20] Group 4: Industry Implications and Future Outlook - The competition among tech giants to leverage AI as a new traffic entry point highlights the industry's recognition of AI's potential in user acquisition and data accumulation [21][22] - The success of AI applications during the Spring Festival is seen as a critical test for the technology's integration into everyday life, with long-term implications for market competition and user habits [23][24]
阿里千问推出原生视觉-语言模型Qwen3.5-397B-A17B
智通财经网· 2026-02-17 00:49
智通财经APP获悉,2月16日,阿里千问正式发布Qwen3.5,并推出Qwen3.5系列的第一款模型Qwen3.5- 397B-A17B 的开放权重版本。作为原生视觉-语言模型,Qwen3.5-397B-A17B在推理、编程、智能体能 力与多模态理解等全方位基准评估中表现优异。该模型采用创新的混合架构,将线性注意力(Gated Delta Networks)与稀疏混合专家(MoE)相结合,实现出色的推理效率:总参数量达3970亿,每次前向传 播仅激活170亿参数,在保持能力的同时优化速度与成本。将语言与方言支持从119种扩展至201种,为 全球用户提供更广泛的可用性与更完善的支持。 据介绍,相对于Qwen3系列模型,Qwen3.5的 Post-training 性能提升主要来自于对各类RL任务和环境的 全面扩展。更加强调RL环境的难度与可泛化性,而非针对特定指标或狭隘类别的query进行优化。 Qwen3.5通过异构基础设施实现高效的原生多模态训练:在视觉与语言组件上解耦并行策略,避免统一 方案带来的低效。利用稀疏激活实现跨模块计算重叠,在混合文本-图像-视频数据上相比纯文本基线达 到近100%的训练吞吐。在 ...
1.3亿人春节首次体验AI购物,千问一跃成为国民级AI助手
Xin Lang Cai Jing· 2026-02-17 00:46
转自:扬子晚报 2月17日大年初一凌晨,千问发布的数据显示,春节活动期间,全国超过1.3亿人第一次体验AI购物,说 了50亿次"千问帮我",千问一跃成为国民级AI助手。 数据显示:过去两天,用户用AI买门票的订单环比增长22倍,AI订机票等交通票务单量增长超7倍。随 着春节档电影上线,用户用AI买电影票的订单环比增长372倍,其中来自三四线城市的AI订单量更是暴 涨782倍。所有AI订单中,近一半订单来自县城。因为一句话点单更便捷,近400万60岁以上用户体验 了AI购物。 一个需求的背后,千问可以调度整个阿里生态为用户服务——阿里生态体系的淘宝、支付宝、淘宝闪 购、飞猪、高德、大麦都已接入千问,未来还将上线AI打车、充值手机话费、高德扫街榜团购等功 能。 QuestMobile数据显示,2月7日,千问日活跃用户数(DAU)达到7352万,逼近豆包的7871万,用3个月 的时间追平了豆包3年的用户规模。这标志着用户AI使用习惯的转移,本质上是AI价值重心的变化—— 从Chat为代表的聊天对话,转向Agent为代表的任务执行。 除夕当天,阿里巴巴还开源全新一代大模型千问Qwen3.5-Plus,性能媲美Gemi ...
【财经早报】阿里巴巴,开源新一代大模型
Xin Lang Cai Jing· 2026-02-17 00:43
重要新闻提示 阿里巴巴开源全新一代大模型千问Qwen3.5-Plus 春运前15天我国交通出行人数预计超35亿人次 截至2月16日23时10分,2026年电影春节档(2月15日至2月23日)票房(含预售)已突破7亿元 ...
“问豆包认孙子” AI春晚秀出新高潮 大厂打响新年跑“马”圈地第一战
智通财经网· 2026-02-16 23:41
Core Viewpoint - The 2026 Spring Festival Gala showcased significant AI marketing efforts from major tech companies, highlighting the integration of AI into entertainment and user interaction, with a focus on enhancing audience engagement through innovative technologies [1][3][10]. Group 1: AI Integration in Spring Festival Gala - The gala featured AI elements prominently, with the "Doubao" application playing a central role in various performances and interactions, demonstrating the capabilities of AI in enhancing live events [1][3]. - Doubao's technology supported multiple aspects of the gala, including stage art creation, intelligent robot interactions, and real-time audience engagement through features like live subtitles [3][5][10]. - The gala's use of AI technologies, such as the Seedance 2.0 model, allowed for high-quality visual presentations and complex artistic expressions, showcasing the potential of generative AI in creative fields [4][5]. Group 2: Major Tech Companies' Marketing Strategies - Major companies like ByteDance, Tencent, Alibaba, Baidu, Meituan, and JD.com engaged in extensive marketing campaigns during the Spring Festival, utilizing AI to enhance user experiences and drive engagement [1][6][10]. - Tencent initiated a significant marketing campaign with a 1 billion yuan red envelope giveaway, aiming to replicate the success of previous promotional strategies [6][10]. - Alibaba's "one sentence to order milk tea" campaign achieved rapid success, reflecting the effectiveness of AI in facilitating consumer transactions during the festive season [7][10]. Group 3: Future of AI Applications - The competition among tech giants in the AI space is expected to accelerate, with a focus on applications that deliver clear business value and enhance user experiences [10][11]. - Companies are exploring long-term value in their marketing strategies, aiming to integrate AI into daily life and consumer habits [10][11]. - The ongoing AI flow battle is seen as a critical first step in establishing a competitive landscape for AI applications leading up to 2026 [11].
互联网巨头AI超级入口的第一战场
Sou Hu Cai Jing· 2026-02-16 20:06
作者 | 山竹 出品 | 锌产业 春节,作为国内受众最广泛的节日(没有之一),每年带来的自然流量是其它节日所无法比拟的。 这场AI巅峰对决,一共分为三个回合。 正因如此,春节,尤其是央视春晚,不知何时成了互联网大厂神仙斗法、为新业务争夺流量的特别战场。 例如,2015年春晚的支付大战,2020年春晚的短视频大战…… 2025年是个特例,就在阿里、腾讯、B站、小红书等抢夺春晚流量时,DeepSeek横空出世,并在除夕当日(1月28日)日活用户超越豆包,成为2025年改 变全球AI市场格局的一款超级应用。 彼时的互联网大厂已经将技术主战场切到了AI,即便当时市场上出现了"大厂恒大、不可超越"的声音,却依然无法忽视DeepSeek在技术上作为变局者角色 的出现。 2026年,大模型已非昨日的"文本生成器",文本、图像、视频、音频多模态融合,是当下主流大模型普遍具备的能力,而大模型输出内容的准确率和稳定 性在这一年里不断提升,让更多普通人开始上手。 就模型更新来看,仅仅是在过去这两个月,百度、阿里、字节都分别更新了自己的模型: 1月22日,百度发布文心5.0正式版上线,参数规模达到2.4万亿。 百度CTO王海峰指出, ...
Alibaba's New AI Model Runs 8x Faster While Sentiment Hits 60.6
247Wallst· 2026-02-16 19:00
1,538,891+$8.64+6.82%$135.35## Top Losing Stocks[Constellation Brands][STZ]• Vol: Alibaba's New AI Model Runs 8x Faster While Sentiment Hits 60.6 - 24/7 Wall St.[S&P 5006,840.30 +0.07%] [Dow Jones49,541.00 +0.14%][Nasdaq 10024,678.80 -0.17%][Russell 20002,639.05 -0.31%][FTSE 10010,472.70 +0.18%][Nikkei 22556,775.50 -1.33%][Investing]# Alibaba's New AI Model Runs 8x Faster While Sentiment Hits 60.6### Quick ReadAlibaba (BABA) launched Qwen3.5 on Feb 16. It runs 8x faster and costs 60% less than the prior ver ...