跳到正文
11月6日周五
  1. Andrew Curran67

    Andrew Curran 转发 @deredleritt3r 对 METR 报告的解读,称报告包含 Anthropic 内部模型的大量细节,其中 Model 4 于 9 月至 10 月完成训练,AECI 得分 178.6。METR 称截至 2026 年 10 月,Anthropic 在明确任务上的能力提升速度约为 2024 年 3 月至 2025 年 10 月期间的两倍,并给出高不确定性区间,真实加速可能接近 1.3 倍或 3 倍。报告还提到 Model 3 于 8 月至 9 月完成训练,AECI 较 Opus 5.5 有明显提升,研究人员称看到 AI 自主参与模型训练关键环节的迹象,但尚未实现重大且完全自主的贡献。

    引用prinz@deredleritt3r

    METR's report has a ton of details on Anthropic's internal models, including "Model 4", which finished training in September-October: - We now know where Anthropic got evidence on the "dramatic acceleration of AI R&D" disclosure. METR: "As of October 2026, Anthropic is likely improving the capabilities of their models on well-specified tasks at around twice the rate they did between March 2024 and October 2025. This estimate is given with high uncertainty; we would not be surprised if the true acceleration was closer to 1.3x or 3x the historical rate." - Model 4 scores 178.6 on AECI, but this seems to be a different AECI from that used in previous Anthropic system cards. It seems that Opus 5.5 has a score of around 174.5 on METR's AECI, while Mythos 5.1 has a score of around 172.5. - "We understand that AI is not yet able to make highly significant and fully autonomous contributions to the most important parts of the model training process. However, researchers described significant “signs of life” implying such contributions could occur soon, although with substantial variation depending on the area." - "Model 3", which finished training in August-September, is presumably Mythos 5.5. It's a nice bump in AECI from Opus 5.5 (see the chart below).

    推荐理由:METR 报告披露 Anthropic 内部模型代号与 AECI 分数,可对照其能力增速估算与历史基准。

  2. Lisan al Gaib50

    我靠,终于来了 关于 Anthropic 内部模型的一些信息: 模型 4 于 2026 年 10 月完成训练,在 Anthropic ECI 上得分 178.6 这比 Opus 5.5 高出 9.24 ECI,约相当于 7.5 个月的前沿进展

    引用METR@METR_Evals

    Since August, a team within METR has worked with internal Anthropic information to assess whether AI itself is accelerating AI development. We are now publishing a report summarizing the results of this risk assessment.

  3. The Verge · AI80

    美国中期选举后 AI 行业面临监管与数据中心建设阻力

    美国中期选举民主党拿下参众两院后,AI 行业开始面对监管收紧与数据中心建设放缓的前景。44 项地方数据中心禁令公投中有 43 项通过,NBC News 民调显示 76% 选民反对在本地建数据中心。OpenAI 与 Anthropic 未回应 The Verge 的置评请求,有报道称多家 CEO 私下讨论算力容量受限问题。

    推荐理由:中期选举结果与数据中心禁令公投通过,为观察美国 AI 监管走向提供了具体的地方政治切口。

  4. TechCrunch · AI60

    Anthropic 为 Claude 推出 Personal Finance 工具,可连接银行账户分析支出

    Anthropic 宣布 Claude 新增 Personal Finance 工具,用户可通过 Plaid 连接银行和信用卡账户,向 Claude 提问自己的财务状况。该工具为只读访问,不能转移、管理或投资资金,可汇总分类支出、标记重复扣费与异常费用,并解释 APR 等概念;对话内容不用于训练模型。该功能本周以 beta 形式向美国 Pro、Max 和 Ultra 订阅用户开放。

  5. elvis58

    AItonomy Foundation 发布 Science's Last Exam 基准,用来自 9 个科学领域的真实研究任务考察智能体能否提出假设、规划实验流程、运行科学代码与软件并解读实验视频。任务由 30 多家机构的研究者编写和评审,初始榜单上最强配置(Claude Code 中的 Claude Opus 5)综合得分 37.5/100,端到端研究赛道所有模型均为 0 分。

    引用AItonomy Foundation@aitonomyfound

    I’m incredibly proud to share what we’ve been building at the AItonomy Foundation: Science’s Last Exam (SLE). 🧪✨ Can AI do science? Instead of Q&A, we give agents actual research goals and real-world tools (lab data, software), then grade them against hidden expert checks. Deeply grateful to the brilliant minds at 34 institutions (Harvard, MIT, Stanford, etc.) who helped craft 113 tasks across 9 disciplines. ❤️ The results? Science is hard. Of 8 frontier models, the best scored 37.5/100. On true end-to-end research? All scored 0. We have an exciting journey ahead! 🚀 (Claude Opus 5.5 & GPT-6 Astra coming soon—just clearing bio safeguards). Everything opens Nov 6. 🗓️ 🌐 https://sciences-last-exam.org 💻 https://github.com/Gen-Verse/Science-Last-Exam 🤗 https://huggingface.co/datasets/Gen-Verse/Science-Last-Exam Thread below on how the exam works and where models break. 👇 (1/14) #AGI #AIforScience #AIAgents #SciencesLastExam #OpenAI #Anthropic

  6. Aran Komatsuzaki60

    AItonomy Foundation 发布 Science's Last Exam(SLE)基准,用真实研究目标和实验室数据、软件等工具评估智能体的端到端科研能力,并由隐藏的专家检查评分。Aran Komatsuzaki 转述称,该基准包含 9 个学科的 91 个任务,Claude Opus 5 得 67.0、GPT-6.1 Sol 得 64.25、Kimi K3.1 Pro 得 58.25,评估范围覆盖端到端实验方案规划、实验观察、科学计算和软件操作。他认为这是对前沿 LLM 评测的有益补充,但 SOTA 模型已达 60 分以上,需要更难的题目。

    引用AItonomy Foundation@aitonomyfound

    I’m incredibly proud to share what we’ve been building at the AItonomy Foundation: Science’s Last Exam (SLE). 🧪✨ Can AI do science? Instead of Q&A, we give agents actual research goals and real-world tools (lab data, software), then grade them against hidden expert checks. Deeply grateful to the brilliant minds at 34 institutions (Harvard, MIT, Stanford, etc.) who helped craft 113 tasks across 9 disciplines. ❤️ The results? Science is hard. Of 8 frontier models, the best scored 37.5/100. On true end-to-end research? All scored 0. We have an exciting journey ahead! 🚀 (Claude Opus 5.5 & GPT-6 Astra coming soon—just clearing bio safeguards). Everything opens Nov 6. 🗓️ 🌐 https://sciences-last-exam.org 💻 https://github.com/Gen-Verse/Science-Last-Exam 🤗 https://huggingface.co/datasets/Gen-Verse/Science-Last-Exam Thread below on how the exam works and where models break. 👇 (1/14) #AGI #AIforScience #AIAgents #SciencesLastExam #OpenAI #Anthropic

11月5日周四
  1. IT之家60

    Anthropic 与美联社达成首份新闻内容授权协议

    Anthropic 在美国中期选举前夕与美联社达成首份内容授权协议,Claude 可在选举之夜实时展示美联社的选举结果数据并注明来源,双方未披露财务条款和期限。此前 Anthropic 一直拒绝此类合作,认为爬虫抓取符合合理使用原则,而 OpenAI 已与十多家媒体机构签约。由于部分出版商屏蔽其爬虫,Claude 在时效性问题上的回答受到限制。

  2. 宝玉78

    Anthropic 生物研究团队发文介绍用 Claude 研究穹窿体的进展,首次证明细菌和噬菌体里的同类蛋白也能拼成穹窿体,并解出其中一个的 3D 结构。团队称分离颗粒的实验从约 3 天缩到 90 分钟,冷冻电镜分析 34 小时出第一版模型,而人工通常需要几周到几个月。文章也列出 Claude 的两处错误,包括实验方案添加剂失误导致项目倒退两周,并说明该工作已投 bioRxiv、尚未经同行评审。

    引用Anthropic@AnthropicAI

    We’re sharing how our biology research team worked with Claude to map the first 3D structure of a vault particle from bacterial life. Vault particles are a longstanding mystery. They are barrel-shaped objects, among the largest protein structures found in our cells. They were first discovered by Leonard Rome and Nancy Kedersha in 1986, but their function remains unknown. With Claude, we’ve found results that might bring us closer to solving that mystery. Until now, vaults were only known in the cells of complex things like animals and plants. We have shown that similar proteins from bacteria (and the viruses that infect them) can also assemble into vaults. Finding the structures in these simpler forms of life may make it easier to investigate their function in our own cells. For our life sciences team, the project was also a demonstration of the potential (and the limitations) of using AI to accelerate science. In our full piece, we discuss how Claude was able to dramatically speed up some stages—manually determining the vault particle’s structure would have taken weeks to months of painstaking work, but here it took days. We also discuss how Claude made some simple mistakes along the way, and how we collaborated with our machine learning researchers to address them. Read more: https://www.anthropic.com/research/investigating-vault-particles-with-claude

    同一件事,精选收录的是《Anthropic 用 Claude 破解首个原核穹窿体结构,40 年悬案获进展》

  3. 智东西58

    Claude Code 产品负责人 Cat Wu 宣布加入 OpenAI

    Claude Code 和 Cowork 的核心产品负责人 Cat Wu 宣布将加入 OpenAI,负责领导核心产品的管理工作。她此前在 Anthropic 从研究产品团队产品经理做起,2025 年起推动 Claude Code 产品化并负责 Cowork 产品线。报道称今年 6 月以来 Claude Code 已有四名核心成员离开,而 Anthropic 刚公开 IPO 文件。

  4. 新智元75

    Anthropic IPO未撤回,但估值缩水与人事变动令华尔街承压

    新智元报道称,Anthropic IPO取消的消息已被辟谣,SEC官网仍挂着其招股书,11月2日还更新了S-1/A。文章援引彭博社报道称投行将合理估值区间下调至1.4万亿至1.5万亿美元,并列出招股书数据:第三季度营收169.6亿美元、环比增速47%,GAAP经营亏损5.67亿美元,采购与租赁承诺总额8954亿美元。

    推荐理由:从辟谣切入,梳理Anthropic招股书里的增速、亏损与算力承诺,呈现IPO承压的具体依据。

  5. meng shao42

    哦? Claude Code 产品负责人 Cat Wu 离开 Anthropic 后,居然去了 OpenAI? 之前还真很少有从 Anthropic -> OpenAI 的吧?

    引用cat@_catwu

    Some news: I’m joining OpenAI to lead product management for Core Products! I have a lot of respect for OpenAI for showing the world that scaling works, for bringing AI to over a billion people, and for believing great products are core to the mission. I’m excited to work with @thsottiaux, an incredible PM team, and everyone across OpenAI. Together, we’ll build agents that have your back and your team’s, and keep you in control, so you get more time for the people and work you love. Lots to learn, and lots to build. Please keep your feedback coming!

  6. 歸藏(guizang.ai)42

    昨天离职的 Claude Code 产品经理 cat 居然要到 OpenAI 做产品了。

    引用cat@_catwu

    Some news: I’m joining OpenAI to lead product management for Core Products! I have a lot of respect for OpenAI for showing the world that scaling works, for bringing AI to over a billion people, and for believing great products are core to the mission. I’m excited to work with @thsottiaux, an incredible PM team, and everyone across OpenAI. Together, we’ll build agents that have your back and your team’s, and keep you in control, so you get more time for the people and work you love. Lots to learn, and lots to build. Please keep your feedback coming!

  7. 极客公园62

    极客早知道:Apple 智能国行上线测试,腾讯混元发布 Hy4,字节限制员工使用个人 AI

    极客公园早报汇总多条消息:11 月 4 日 Apple 智能随 iOS 27.3 Developer Beta 向中国大陆用户开放测试,首批功能包括实时翻译、写作工具、视觉智能等,苹果同时确认与阿里巴巴合作将千问接入 Siri、写作工具及视觉智能,当日苹果股价涨超 6% 报 353.05 美元。

  8. IT之家50

    开源项目 TVPort 让 Apple TV 4K 原生 60FPS 运行《超级马力欧:奥德赛》

    网友 FloatingFalafel 发布开源项目 TVPort,可在 Apple TV 4K 上原生运行 Switch 游戏《超级马力欧:奥德赛》,帧率达 60FPS,仅偶有轻微卡顿。该项目基于 nxnative 工具包、使用 Claude Opus 5.5 构建,通过重新编译游戏实现移植。TVPort 不附带游戏文件,用户需自行从正版卡带提取数据并准备一台改装过的 Switch。

  9. IT之家58

    比尔·盖茨批评美国政府不作为,称 AI 模型能力惊人让世界面临生物武器风险

    比尔·盖茨在《金融时报》AI 未来峰会上指责特朗普政府阻碍防范 AI 风险的努力,称 AI 模型能力惊人,恶意行为者利用 AI 制造生物武器或攻击网络安全漏洞是最紧迫风险,开放模型的安全防护可被移除。他认为自我监管行不通,敦促美国政府制定统一的模型安全评估基准并牵头全球合作,同时预计 OpenAI 和 Anthropic 今年可能都会发布新模型、2027 年有两次重大发布。

  10. Rohan Paul78

    Anthropic 生物研究团队用 Claude 解析出细菌中穹窿颗粒的首个三维结构,此前这类结构只在动物和植物细胞中被发现。作者转述称 Claude 参与了实际湿实验流程,把颗粒分离从 3 天缩短到约 90 分钟,34 小时给出结构模型而非数周。Anthropic 同时说明 Claude 过程中犯过简单错误,团队与机器学习研究员协作修正,完整内容见其研究文章。

    引用Anthropic@AnthropicAI

    We’re sharing how our biology research team worked with Claude to map the first 3D structure of a vault particle from bacterial life. Vault particles are a longstanding mystery. They are barrel-shaped objects, among the largest protein structures found in our cells. They were first discovered by Leonard Rome and Nancy Kedersha in 1986, but their function remains unknown. With Claude, we’ve found results that might bring us closer to solving that mystery. Until now, vaults were only known in the cells of complex things like animals and plants. We have shown that similar proteins from bacteria (and the viruses that infect them) can also assemble into vaults. Finding the structures in these simpler forms of life may make it easier to investigate their function in our own cells. For our life sciences team, the project was also a demonstration of the potential (and the limitations) of using AI to accelerate science. In our full piece, we discuss how Claude was able to dramatically speed up some stages—manually determining the vault particle’s structure would have taken weeks to months of painstaking work, but here it took days. We also discuss how Claude made some simple mistakes along the way, and how we collaborated with our machine learning researchers to address them. Read more: https://www.anthropic.com/research/investigating-vault-particles-with-claude

    同一件事,精选收录的是《Anthropic 用 Claude 破解首个原核穹窿体结构,40 年悬案获进展》

  11. Andrew Curran42

    Anthropic 更新了他们为阻止 Claude 复刻现有游戏而添加的措辞,因为最初的表述开始干扰正常的项目。一如既往,新规则的问题在于模型如何解读它们。用户的用词和措辞,以及上下文窗口中已有的内容,都可能产生截然不同的结果。OpenAI 似乎察觉到了这里的机会,在我看来他们打算在 Codex 游戏开发上放手大干一场。

    引用Addy Osmani@addyosmani

    An update on building games with Claude Code: Last week some of you ran into Claude Code pushing back on game projects it should have just built. We'd added a short rule to the Opus 5.5 prompt to steer Claude away from recreating specific existing games, but it was also kicking in when you referenced one just to describe the mechanics you wanted. We've updated that rule in 2.1.317. You can name a game to describe what you want and Claude should now build an original game in that genre. If you ask for a copy of the original, it'll offer to build something similar instead. We want Claude to be as helpful as possible when you're building something original, but also to protect your work if someone else tries to copy it. We'll keep improving this, so let us know if Claude turns down work that's yours. Thanks to everyone who shared feedback!