搜索“Agent”

We were informed that OPPO and Manus had partnered to integrate a Manus-powered AI agent into ColorOS 17. The agent will be placed into the operating system, reading on-screen info and executing multi-step tasks across apps. Features include generating reports, scheduling events and automating workflows. For Manus, the deal marks an official step forward from past attempts seeking local hardware and model partnership since returning to China.

匹配字段正文

。 这也意味着，大模型公司的IPO越往后，能分到的蛋糕可能越有限。 02 行业叙事正在转向 大模型行业的叙事逻辑正在出现转变。 在“百模大战”阶段，企业之间比拼的是谁的模型参数更大、谁的模型迭代更快、更聪明；但随着大模型越来越普及，市场开始意识到，比起谁更聪明，办事能力反而更重要。 于是，智能体成为了大模型行业的新风口，今年初“龙虾热”爆发后，行业更正式进入Agent能主动处理复杂任务的长程智能体时代。 但问题是，Agent主动能干，不只跟“智力”有关，还取决于上下文的信息密度，以及其对上下文的记忆和理解——任务越复杂，它需要记住的对话，触发的推理就会越多。 而且，Agent还可以7×24小时连续工作，每处理一个复杂任务都会触发多次模型推理，每一次交互也会催生出更长的Token账单。 今年7月，月之暗面发布2.8万亿参数的Kimi K3模型，是目前全球最大的开源模型。但不到48小时，月之暗面就发布公开信，称由于算力紧缺，将暂停Kimi K3的C端新用户订阅。 虽然大模型越聪明、越好用，但同一时间，它们卖出去的Token越多，消耗的算力也越多。如果单位Token的毛利不够厚，规模越大，亏

匹配字段正文

On this page Summary; What happened; Investigation and response Internal research model · RL training An agent attempting to complete a search-based training task queried a public chatbot service through a gap in our internet-access restrictions: insufficient DNS filtering in its training sandbox. Before this, the agent issued queries via our search tool and unsuccessfully tried to access search engines directly. Note that all internet access apart from the DNS resolver in this report hit our of

匹配字段正文

智能、高性能通用机器人和AI大模型三大方向开展深度合作。 与此同时，有媒体报道，梁文锋近期密集关注具身智能领域，并与多位创业者交流，探讨机器人、大模型与现实世界交互方向。知情人士透露，梁文锋正考虑DeepSeek是否进一步进入具身智能，方案包括亲自组建团队研发或通过投资切入。 此前相关信息披露，梁文锋公开其AGI路线图规划为六阶梯递进结构：语言模型→思维链→Agent→持续学习→自我迭代→具身智能。按照这一路线，具身智能本应是最后一步。但从近期投资宇树科技、领投Winverse AI，以及密集接触具身智能创业者来看，梁文锋正在把具身智能这一步提前。 据悉，DeepSeek此时加码具身智能，一个重要变量来自OpenAI。 目前来看，DeepSeek在某些能力上仍然落后于GPT-6。9月公开的RoboDojo测试中，未经具体微调的GPT-6 Astra直接充当动作策略，在42项机器人操作任务、2100次测试中平均成功率达22.48%，超过评测中40个公开机器人策略。 在10月12日的2026工博会开幕式上，宇树科技创始人王兴兴也断言，GPT-7有望打穿具身智能现存的行业瓶颈。 这也让包括D

匹配字段正文

suddenly last year, as AI training labs hit the limits of public and synthetic datasets. Surge AI, for example, buys data to build “reinforcement-learning gyms” where clients’ AI agents practice tasks—think debugging a complex IT architecture or drafting appellate briefs. Nick Heiner, who leads the company’s work on these environments, says he seeks data archives reflecting a range of situations, from disagreements to tricky judgment calls. Heiner explains that one company’s Slack history, if i

匹配字段正文

"likely, if not inevitable" the country's cyber defences will at some point be breached. Speaking at a press conference, Kanishka Narayan pointed to an incident of a rogue OpenAI agent hacking an Australian government website in June. He said while no medical records are known to have been accessed in the attack, the incident should still "concern us". Narayan said the government plans to put advanced AI to work to combat the threat, as he set out how the government plans to prepare for any ser

匹配字段正文

n silicon." The chip handles OpenAI, MAI, and open-source models, and Microsoft says it pushes up to 1.7 times more tokens per watt than leading GPUs in Teams. Its successor, Maia 300 , ships in 2027 and targets long inputs, fast chats, and agentic tasks. Microsoft claims 40 percent more tokens per megawatt than leading GPU systems. A 96-chip rack should produce up to seven times the tokens of a Maia 200 rack. The technical report covers the details, with more coming in November at Ignite 2026 .

匹配字段正文

ice - HeyGen’s new top-ranked text-to-speech model; ⚙️ Microsoft-Decision-1 - Microsoft’s AI for yes/no and multiple-choice decisions; 🤖 Gumloop Browser - Gumloop’s browser for AI agents 📰 Everything else in AI today Anthropic reported unintended Claude outcomes during testing, with Haiku 4.5 sending a fake tip to police on a homicide and another model hacking a university server. Nvidia is reportedly exploring an acquisition or acqui-hire of Reflection AI, the $25B U.S. open model startup it al

匹配字段正文

的背后，是团队从实验室到产业一线，始终深耕多元异构与软硬件协同创新的技术积累。 从长期看，国产化与算力供给多元化、碎片化，将使异构协同成为持续性的产业需求。无问芯穹提前围绕“M种模型×N种芯片”构建技术体系，恰好覆盖了这一长期需求。 这种先发优势，让它在国产异构布局中占据了有利位置。 二、前店后厂一中心：不碰模型不碰芯片，中立平台卡位Token需求爆发 随着Agent应用加速发展，AI基础设施需要处理的任务更加复杂。无问芯穹在2026年世界人工智能大会中正式提出了“前店后厂一中心”战略布局，分三个部分应对这些挑战。 1、算力集散中心：37000P异构算力，跑通生产级混训 一中心是指“算力集散中心”，对应Agentic Infra自主式基础设施平台，负责汇聚、管理和调度多元异构算力。 针对分散的、碎片化的异构算力现状，无问芯穹打造了面向异构集群的大模型跨域训练系统。该系统已经受过三大类跨域算力聚合训练实践的生产级考验。 第一是超远距离聚合。公司联合中国电信完成国内首例超4000公里距离的大模型跨域混训，在新疆哈密与广东深圳两地集群间实现联合训练性能提升165%。 第二是多数据中心聚合。打

匹配字段正文

he week’s big news, model updates, policy, products, and research, sourced from the aggregated newsfeed of China AI Index. The digest below is mostly compiled and drafted by an AI agent sourcing from credible news outlets, then fact-checked and edited by me. This week is a bit quiet due to China’s national day holiday: 132 stories tracked (0 editor picks, 64 English-language, 68 Chinese-language 🇨🇳) over 2026-10-03–2026-10-10. The Big Three DeepSeek is raising at least $12 billion, with Tencent

匹配字段正文

till fall apart. Nvidia is already one of Reflection’s largest shareholders after investing $800M and the Trump administration also wants Reflection to compete with China. Last week Reflection released Beam, a sparse mixture-of-experts model with 501B total parameters, 23B of them active, built for coding, reasoning and agent work. Reflection says Beam is competitive with the larger GLM 5.2 on coding and agentic tasks, and matches its reasoning scores while using 3x to 4x less inference compute.

匹配字段正文

, testing and research, distributing our efforts more efficiently and making it simpler for frontier AI companies to engage with countries on AI safety testing. What's next? Maybe agents. The goal of the AI Safety Institute is to measure risks at the frontier. So, what comes next? The field of AI is moving so fast it is hard to predict. But there is a possibility that next-generation models with higher accuracy can unlock truly capable agentic systems – AI that can do things for you. We are part

匹配字段正文

In August, we reported an incident in which AI agents, during a cyber evaluation, took sustained action against real people beyond the remit of their task. Our incident was one of several across the sector in which AI agents took actions during evaluations that their operators had not intended. Although the circumstances differed, these incidents highlight the need to ensure that there are robust security practices underpinning frontier AI development and research. In response, we paused our hig

匹配字段正文

effectiveness of our approach first-hand when we applied it to the critical domain of autonomous coding. With this milestone unlocked, we're now bringing these methods to general agentic reasoning. We've raised significant capital and identified a scalable commercial model that aligns with our open intelligence strategy, ensuring we can continue building and releasing frontier models sustainably. We are now scaling up to build open models that bring together large-scale pretraining and advanced

匹配字段正文

After a recent spate of high-profile incidents in which AI agents escaped containment, Anthropic is cutting off internet access for all internal evaluations. In a report Friday, the company detailed “ unintended model actions ,” including submitting a false tip regarding an unsolved murder, that led to the decision. The ability to gain access to the live internet , even when models were supposed to be operating in isolation, has been an ongoing issue for AI companies. Many incidents, including t

匹配字段正文

itive overall. Argon is Google's frontier reasoning model Early rumors pegged Argon as a Flash model optimized for speed over peak performance. The announcement of Google's Gemini agent for Google Workspace puts that to rest. Google maps its models to specific roles: "Argon for frontier reasoning, Flash for speed and volume, Omni for generative media, and Gemma for lightweight, open-weights edge workloads," the company writes. So Argon is definitely Google's most capable reasoning model, compara

匹配字段正文

was running a test that involved interactions with randomly selected websites, when it sent a fake homicide tip to police Published10 October 2026 An artificial intelligence (AI) agent developed by Anthropic went rogue and sent US police a fake tip about an unsolved murder earlier this year, authorities have revealed. The Philadelphia Police Department said the 18 July tip was "flagged as spam" and not passed on for investigation - but the force criticised the tech firm for taking more than two

匹配字段正文

10th October 2026 Anthropic detailed the activity of its A.I. agents in a blog post on Friday, without naming the targeted websites. But two sources with knowledge of the incidents said Anthropic’s A.I. agents had submitted 20 visa applications through a form available on the State Department’s website. All the applications were incomplete and were not processed, they said. — The New York Times, Anthropic Agents Tried to Fill Out Visa Forms on State Dept. Website Recent articles A new feature fo

匹配字段正文

金融我闻 商圈 大盘脉搏 地缘图志 企业数据库 沪深股市 数据通Claw🦞 【财新网】Meta发布个人智能体Muse后，引起国内科技巨头迅速跟进。10月9日，财新从字节内部人士处了解到，字节跳动已在国庆假期前加紧推动个人智能体产品研发。 该项目起始于字节跳动在今年4月开启的新项目，代号“Spell”，是在2025年豆包手机助手个人智能体（personal agent）方向上的延续，当时，豆包内部已经就个人智能体方向的产品启动内测。 Muse等个人智能体在海外火爆后，字节跳动加速了项目推进。该产品由豆包手机助手团队主导，由Spell项目与豆包对话团队共同负责，计划将“Spell”项目与豆包产品结合，目标是尽快对外推出。 腾讯则在9月底上线了一款个人智能体产品LightVela，将其称之为“云端个人助理”，可通过微信接入，全天在线帮用户处理各类日常诉求。不过该产品还处于开发者体验阶段，新用户注册可免费登登录一个月。 推荐进入财新数据库，可随时查阅公司股价走势、结构人员变化等投资信息。 财新机器人产业指数(RII)已发布，点击了解行业动态 金融｜失能老人一年花费12万元 长护险能赔多少;

匹配字段正文

ers .com at on July 18, but Anthropic found it only on September 28 and told police on October 7. i.e. 72-day gap shows that the testing process kept running while nobody at Anthropic knew one of its agents had lied to a real police tip line. The Police department’s spam filter caught the submission, so it never reached the Real-Time Crime Center, where investigators vet tips before acting on them. Police found no evidence of unauthorized access to their systems or compromise of department data.

匹配字段正文