此次开源意味着硬件厂商、手机厂商和开发者均可基于AutoGLM，在自己的设备或系统中复现一个 能"看懂"屏幕、并模拟真人进行点击、输入、滑动的AI助手。目前，AutoGLM已支持微信、淘宝、抖 音、美团等超过50个高频中文应用的核心场景，其自动化操作能力与此前引发热议的"豆包手机"演示相 似。 智谱深夜开源其核心AI Agent模型AutoGLM。该模型被业界视为全球首个具备"Phone Use"(手机操作)能 力的AI Agent，能够稳定完成外卖点单、机票预订等长达数十步的复杂操作流程。 AutoGLM的开源将大幅降低AI手机的技术门槛，推动AI手机生态从封闭走向开放共创。同时，项目支 持本地与云端部署，确保数据与隐私控制权始终掌握在使用者手中。 ...

Core Insights - The article discusses the rapid success and technological foundation of the "Doubao Phone" and its assistant, which has gained significant attention in the market due to its advanced capabilities in automating tasks on mobile devices [1][50]. Group 1: Product Overview - The "Doubao Phone" sold out its initial stock of 30,000 units, with prices in the second-hand market doubling [1]. - The phone's assistant can automate complex tasks across applications, such as submitting leave requests and booking train tickets [4][5]. - The assistant is built on ByteDance's self-developed UI-TARS model, which has been optimized for mobile use [7][8]. Group 2: Technological Development - The UI-TARS model has undergone significant iterations, with the initial version released in January 2023, followed by UI-TARS-1.5 and the latest UI-TARS-2, which enhances the agent's capabilities [11][23][34]. - UI-TARS-2 addresses issues related to data scalability and multi-round reinforcement learning, allowing for more autonomous interactions with graphical user interfaces [34][35]. - The model has shown superior performance in various benchmarks compared to competitors like OpenAI's models [27][28]. Group 3: User Experience and Feedback - Users have reported high satisfaction with the assistant's ability to perform tasks efficiently, with one user describing it as the "world's first true AI smartphone" [69]. - The assistant's design includes a dual-mode system, allowing for both rapid responses and deeper reasoning capabilities [60][62]. - Concerns regarding privacy and security have been raised, but the company has emphasized that user consent is required for high-level permissions [50][51]. Group 4: Market Implications - The success of the "Doubao Phone" indicates a shift towards AI-driven mobile technology, where devices can autonomously understand and execute user intentions [85]. - The product's development reflects a broader trend in the industry towards integrating advanced AI capabilities into everyday technology, potentially redefining user interaction with mobile devices [86].

智通财经记者 | 宋佳楠 12月9日，智通财经获悉，"大模型六小虎"之一的智谱AI宣布开源核心AI Agent模型AutoGLM，称希望 通过这款模型，让AI真正学会"使用手机"。 谈及在当下这个时间节点选择开源的原因，智谱表示，"这件事（AI手机）只在一家公司做，是不够 的。"智谱认为，AI手机已经是趋势，但如果"会用手机的AI能力"只掌握在极少数厂商手里，那便意味 着两件事：一是开发者的创新空间，要看平台愿不愿意开放接口；二是用户日常生活中最重要的那台设 备，越来越像"别人家的入口"。 去年10月，智谱发布了首个可在真机上完成一条完整操作链路的AutoGLM。一个月后，AutoGLM发出 了由AI完成的手机红包。 进入2025年，该公司又发布了AutoGLM 2.0，并把它放进一台和用户真实世界脱离的虚拟手机里。基于 此，这台手机跑在云端，每一个动作都可回放、审计、干预，也可以对敏感数据加以隔离。 智谱特别提到，"让AI会用手机之前，我们要先保证，它不会在不该点的地方胡乱伸手。"为此，其主动 放弃了操作微信等这些离用户隐私较近的APP。 但智谱强调，Phone Use能力一旦做大，天然会碰到敏感信息，包括 ...

作者 ｜ 木子、高允毅 2023 年 4 月，在很多人刚听说"大模型"这个词的时候， 智谱团队 开始研究一个听上去不太现实的目标： 让 AI 真正学会"使用手机" ，也就是像真人一样，对智能手机等设备具有使用能力。 32 个月后，智谱把阶段性 重要成果、核心 AI Agent 模型：AutoGLM 给 完整开源 了出来，并放话："每台手机，都可以成为 AI 手机。" 目前，AutoGLM 已支持上百个主流 APP，包括以下这些： 另外，AutoGLM 还能同时在上千台云端虚拟手机里"练功"，通过强化学习等极大地扩展了 Agent 的准确性和泛化能力。而且它被严格关在虚拟设备的安 全沙箱里，既能自由试错，又不会碰到用户真实手机上的隐私数据。 智谱今天开源的是一整套可以"拿来就用"的能力，具体包括： 模型会以 MIT 开源许可证 的形式开放，而所有代码会以 Apache-2.0 开源许可证 的形式，托管在 GitHub 仓库中：github.com/zai-org/Open- AutoGLM。 为什么选择开源？ "从产品的角度，AutoGLM 已经可以支撑起很多真实场景；从工程的角度，AutoGLM 的积累足够写 ...

当前产业相关方关于"豆包手机"的系统安全、信息隐私&合规、生态主导权争夺、利益分配不明等方面 仍存在分歧与争议，虽然这一款搭载豆包手机助手的新机并不完美，但仍值得高度重视。参考汽车产 业，自动驾驶落地过程中面临了比当前AI手机更多的质疑和困难，但在远期光明前景驱动下已成为各 家车企核心战略投入方向；对于手机产业也有类似结论，AI手机有望为手机产业带来更广阔的发展空 间，同时也有可能改变商业模式、价值链以及生态格局，豆包的入局意味着各大手机厂商都必须进一步 加快拥抱AI的节奏以取得更有利的竞争态势。此外，中信证券认为"豆包+努比亚"的合作可类比汽车领 域"华为+赛力斯"的成功案例，大模型厂商有望为手机厂商提供更强大的大模型能力+生态助力，进而 加速能进一步优化消费者体验的AI手机爆款产品加速落地。 本文来自格隆汇专栏：中信证券研究 作者：徐涛 胡叶倩雯 夏胤磊 梁楠 AI手机是手机大厂高度重视的下一个重大变革，豆包等三方大模型厂商看重手机在端侧AI时代作为核 心硬件载体亦积极入局，2025年底豆包与努比亚联合发布搭载豆包手机助手的AI手机产品，具备系统 级智能体+高权限+跨应用执行能力，中信证券认为其重大意义可 ...

Core Insights - The core viewpoint of the articles is that the open-sourcing of the AutoGLM AI Agent model by Zhipu will significantly lower the technical barriers for AI mobile applications, transitioning the AI mobile ecosystem from a closed to an open collaborative environment [1][2] Group 1: AutoGLM Model Features - AutoGLM supports both local and cloud deployment, ensuring that data and privacy control remain with the users [1] - The model is recognized as the world's first AI Agent capable of "Phone Use," enabling it to perform complex operations like food ordering and flight booking through a stable operation chain [1] - The upcoming AutoGLM 2.0 version will enhance the model's learning and generalization capabilities across thousands of virtual device environments [1] Group 2: Industry Impact - The open-source initiative allows hardware manufacturers, mobile phone companies, and developers to create AI assistants that can understand screens and simulate human interactions [2] - Following the announcement, the A-share market saw active performance in AI mobile and PCB sectors, with companies like Shenghong Technology and Ultrasonic Electronics experiencing significant stock price increases [2]

智谱表示，此次开源的是一整套可以"拿来就用"的能力，而不仅仅是一份概念说明。具体包括训练好的 核心模型；Phone Use能力框架与工具链；可直接跑通的Demo，覆盖超50个高频中文App；针对Android 的适配层与示例工程；文档、快速上手指南等。 今年8月，智谱宣布发布全球首个手机Agent——AutoGLM 2.0，由国产模型GLM-4.5与GLM-4.5V驱动， 具备推理、代码和多模态处理能力，能够在多种设备和场景中运行，帮助用户执行操作。 传导至二级市场上，12月9日，AI手机概念短线拉升，福蓉科技（603327）涨停，胜宏科技 （300476）、道明光学（002632）、中兴通讯（000063）等个股纷纷走高。同时，电广传媒 （000917）、奥飞数据（300738）等智谱AI概念也陆续上扬。 12月9日，智谱深夜开源其核心AI Agent模型AutoGLM。该模型被业界视为全球首个具备"Phone Use"（手机操作）能力的AI Agent，能够稳定完成外卖点单、机票预订等长达数十步的复杂操作流程。 据了解，此次开源意味着硬件厂商、手机厂商和开发者均可基于AutoGLM，在自己的设备或系统中复 ...

凤凰网科技讯 12月9日，努比亚总裁倪飞近日就公司新品"努比亚M153豆包手机助手技术预览版"回应 外界关注。倪飞表示，在手机行业缺乏颠覆式创新的背景下，AI手机是明确的发展方向，努比亚选择 以开放姿态与豆包手机助手合作，旨在提升用户体验。 倪飞提到，努比亚早在2017年就开始布局AI，并坚持"AI for All"理念，致力于将AI技术普及化。从首款 AI手机Z17到2025年多模型协同理念，再到与豆包手机助手合作的M153，努比亚称其近十年来持续推 进AI技术创新与全场景布局。 对于目前产品反馈的问题，倪飞表示正与合作伙伴积极沟通解决。努比亚方面称未来将继续专注于用户 体验相关的技术创新，并与行业及用户共同推动"AI for All"进程。 ...

Core Insights - The article discusses the open-sourcing of the AutoGLM AI Agent model by Zhipu, which is set to be released in October 2024 and is considered the world's first AI agent capable of "Phone Use" [1] - AutoGLM can perform complex operations such as food ordering and flight booking, involving dozens of steps, and has been upgraded to version 2.0 as of August 20 this year [1] Group 1 - The open-sourcing of AutoGLM allows hardware manufacturers, mobile phone companies, and developers to replicate an AI assistant that can "understand" screens and simulate human actions like clicking, inputting, and sliding [1] - AutoGLM currently supports over 50 high-frequency Chinese applications, including WeChat, Taobao, Douyin, and Meituan, showcasing its automation capabilities similar to the previously discussed "Doubao AI mobile assistant" [1] - Zhipu claims that the open-sourcing will significantly lower the technical barriers for AI phones, promoting an open and collaborative AI phone ecosystem [1] Group 2 - The project supports both local and cloud deployment, ensuring that data and privacy control remain in the hands of the users [1]

Core Insights - The emergence of AI smartphones is facing significant challenges, particularly highlighted by the collaboration between ByteDance and ZTE, which encountered operational issues due to system-level permission conflicts with major applications like WeChat [1][4][5] - The competition for AI integration in mobile devices is intensifying, with questions about how new entrants can gain trust within established ecosystems dominated by giants like Tencent [2][6] Technical Challenges - The AI assistant, Doubao, attempted to utilize system-level permissions, allowing it to simulate user actions across applications, which triggered security alerts from popular apps [4][5] - This capability effectively positioned Doubao as an invasive entity within the mobile ecosystem, raising concerns about user data privacy and security [5][6] Ecosystem Dynamics - The conflict is not merely technical but rooted in the competitive dynamics between ByteDance and Tencent, where trust and data security are paramount [6][9] - Doubao's intent to streamline user interactions across multiple applications poses a direct threat to the existing flow of traffic and user engagement for established platforms like WeChat and Alibaba [7][8] Strategic Implications for Tencent - Tencent is likely to adopt a more integrated approach by leveraging its existing ecosystem, potentially creating an AI assistant that seamlessly integrates with WeChat, thus enhancing user experience without compromising security [13][14] - The strategy may involve deep partnerships with manufacturers that lack strong ecosystems, allowing Tencent to maintain control over the software and user interactions [11][12] Market Reactions and Future Outlook - The response from existing applications to Tencent's potential AI assistant will be critical, as many rely heavily on WeChat for user engagement and transactions [21][22] - Tencent's approach may involve a combination of incentives and pressures to ensure cooperation from third-party applications, thereby reinforcing its ecosystem dominance [23][24] - The future of AI smartphones may hinge on whether they can navigate existing ecosystem barriers or if they will need to align with established players to succeed [26]