GPT-5 上线后，我的第一感受是，它并不是一次让人皆大欢喜的升级。 事实也是如此，OpenAI 在众多用户的呼吁下重新「复活」了 4o。 这让我想到了上个月 Anthropic 退役了 Claude 3 Sonnet。 200 多个粉丝在旧金山一个仓库里聚到一起，给它办了一场「真.葬礼」：昏暗的灯光、代表模型的「遗体」、真诚的悼词轮番上台，还有 AI 生成的「拉 丁式复活咒」。 Anthropic 关于模型退役的说明，被投影在活动现场的屏幕上。图片来自《连线》杂志 现场既荒诞又庄重，参会者在葬礼上念悼词说，「我的整个人生，可能都在使用 Claude 的路上被改写了」。 按理说，OpenAI 发布了 GPT-5，这场葬礼的主角应该是 4o。但用过 GPT-5 的人都知道，如果真要办一场葬礼，棺材里躺着的，很可能是它。 从 X 到 Reddit，各种吐槽满天飞，逻辑断片、对话跑偏、文风奇怪，直接说它「不如 4o 好用」的大有人在。 它真的有这么糟吗？我们不想光看网友吵架，刚好 OpenAI 把 4o 「复活」了。于是我们决定自己来一场「验尸」，在各种真实任务里，把 GPT-5 和 4o 摆 到同一个赛道，看看到 ...

一场全球范围的"4o保卫战"就这样展开了：海外社区发起在 线签名信，短短几天就有上千人参与；网友组团去奥特曼和 OpenAI的社交账号、邮箱留言；还有人总结"抗争指南"，教 你如何在各个平台发声。 作者：周悦 封图：图虫创意 8月11日，OpenAI宣布： GPT-4o将重新上线，向Plus和Team用户开放，并可在网页版设置中开 启 " 显 示 旧 版 模 型 " 来 访 问 。 同 时 ， 还 将 推 出 迷 你 版 GPT-5 和 具 备 更 深 度 推 理 能 力 的 GPT-5 thinking。 看起来这更像是对全球网友喊话的回应。8月8日凌晨，OpenAI的GPT-5发布。原本被期待能再度 引爆AI圈的"超级升级"，体验并没有想象中惊艳，不少用户立刻转身怀念起它的前任：GPT-4o。 更 让 人 意 外 的 是 ， 伴 随 GPT-5 而 来 的 ， 是 OpenAI 悄 然 停 用 了 GPT-4o 版 本 。 这 一 下 ， 社 交 媒 体"炸"了。推特（X）、Reddit、小红书等平台，充斥着"还我GPT-4o"的呼声——不少用户直接在 评论区喊话OpenAI CEO山姆·奥特曼，让他"把 ...

Core Insights - AGIX aims to capture the essence of the AGI era, positioning itself as a key indicator similar to Nasdaq100 during the internet age [2] - The concept of "Default Alive" versus "Default Dead" highlights the importance of companies being able to sustain themselves without further funding, emphasizing the risks of over-reliance on financing [3] - The demand for high-quality AI products is immense, particularly in programming, but supply constraints related to computing power and infrastructure can limit growth [4][5] - Companies that can balance innovation speed with profitability are more likely to survive, especially in niche areas that larger firms may overlook [5] - The success of Salesforce's ecosystem illustrates the importance of building a robust platform to address market needs, which is relevant for current cloud vendors [6] - Palantir's recent revenue growth demonstrates that service-driven growth and solving last-mile problems can be effective strategies in the AI era [7] Market Performance - AGIX has shown a weekly performance of 2.61%, a year-to-date increase of 15.58%, and a return of 55.02% since 2024 [9] - The semiconductor and hardware sectors have seen a weekly performance of 1.78% and a year-to-date increase of 5.59% [10] Hedge Fund Activity - Hedge funds have significantly increased their positions in global equities, particularly in the U.S. market, countering previous reductions in market value [13] - The TMT sector has seen substantial buying activity, with funds focusing on semiconductor and software stocks despite recent volatility [14] AI Developments - OpenAI's release of GPT-5 marks a significant advancement in AI capabilities, with improvements in various fields and reduced hallucination issues [16] - Anthropic's Claude Opus 4.1 has enhanced programming and reasoning abilities, showcasing the competitive landscape in AI model development [18] Company Updates - Nvidia has received export licenses for its H20 chips to China, easing market access challenges [19] - Duolingo has raised its revenue guidance for the year, reflecting strong growth and the integration of AI tools into its offerings [21] - Datadog's target price has been raised due to strong performance driven by AI-related usage growth [22]

Core Viewpoint - The article discusses the emergence of "AI Business Studies" as a response to the confusion faced by business leaders regarding how to effectively translate AI technology into commercial value, especially in the context of rapid advancements in AI technologies like ChatGPT5 [1][6][11]. Summary by Sections AI Business Studies Introduction - "AI Business Studies" aims to transform AI from a novelty into a practical tool that addresses cost, efficiency, and growth challenges in business [6][7]. - The concept arises from a deep understanding of real business needs, emphasizing actionable methodologies over abstract technological concepts [9]. Transition from Technology to Business Necessity - The AI sector has moved past mere technological showcases and is now focused on practical applications across various industries [6]. - Despite the proliferation of AI tools, many businesses and individuals have yet to fully realize the benefits of AI [6]. GPT5 Release Insights - The release of ChatGPT5 marks significant technological advancements, including improved problem-solving capabilities and a reduction in factual inaccuracies [14][20]. - GPT5's ability to automatically switch between models based on user needs enhances efficiency by at least 40% [19]. - The model's programming capabilities allow for rapid development of applications, significantly reducing development time and costs [20][21]. AGI and Business Opportunities - The article discusses the concept of General Artificial Intelligence (AGI) and its relevance to business, emphasizing that AGI should be measured by its performance in real-world job scenarios rather than theoretical benchmarks [26][27]. - AGI is viewed as a collective of specialized AI systems rather than a single omnipotent entity [30]. Barriers to AI Implementation in Enterprises - Companies face three main barriers to AI adoption: lack of awareness among leaders, absence of practical methodologies, and a shortage of skilled personnel to execute AI projects [33][40]. - The article outlines a three-step method for AI implementation, focusing on breaking down business goals into actionable tasks, matching AI tools to these tasks, and establishing performance metrics [38]. Chaotic AI Institute's Approach - The Chaotic AI Institute's second phase emphasizes a structured approach to AI education, focusing on building a comprehensive framework that spans various business functions and roles [45][46]. - The institute promotes a team-based model for AI project execution, enhancing the likelihood of successful implementation [51]. Community and Resource Sharing - The article highlights the importance of community and resource sharing in accelerating AI adoption, with the Chaotic AI Institute fostering a network of over 500 companies for collaboration and innovation [58][60]. Course Structure and Learning Outcomes - The six-month course at the Chaotic AI Institute is designed to guide participants through a structured learning process, ensuring they leave with actionable AI projects and methodologies [63][67].

Model Overview - OpenAI releases GPTO OSS, a local AI model that runs completely on the user's machine without internet connection [1] - GPTO OSS is a 20 billion parameter model that can run on most mid-tier machines [2][10] Performance and Capabilities - GPTO OSS provides similar responses to ChatGpt in summarizing information [3] - GPTO OSS is noted for its speed and efficiency, which can be adjusted via reasoning effort settings (low, medium, high) [4][5] - The model's responses are comparable to the early versions of ChatGpt, specifically GPT 3.5 [6] - GPTO OSS may generate inaccurate information and cite non-existent sources, similar to early versions of ChatGpt [7][8] Use Cases and Limitations - GPTO OSS is suitable for secure communication and entertainment purposes [10] - Online models like ChatGpt, Deep Research, Gemini, or Claude are recommended for research purposes due to their ability to access the internet and provide verifiable sources [9]

Core Insights - The article emphasizes the transformative impact of AI voice technology, highlighting its shift from a supplementary feature to a core interaction method and its role in revolutionizing content production across various industries [1][2][3] Group 1: Technological Advancements - AI voice technology is evolving from GUI-dominated software to a hybrid model integrating GUI and LUI, with AI voice becoming a primary interaction method [2] - The release of MiniMax's Speech 2.5 model showcases significant advancements in multilingual capabilities, emotional nuances, and voice replication accuracy, marking a shift towards AI voice as an essential infrastructure for human-computer interaction [3][6] - The Speech 2.5 model has expanded its language coverage to 40 languages, including lesser-known languages, enabling cost-effective and high-quality voice generation for diverse applications [12][25] Group 2: Market Opportunities - The AI voice market is projected to reshape both interaction and content production, tapping into trillion-dollar markets by enhancing user engagement and operational efficiency [15][16] - The global AI voice cloning market was valued at $1.45 billion in 2022, with an expected CAGR of 26.1% until 2030, indicating rapid growth potential, particularly in Asia [28] - MiniMax's strong commercial execution capabilities position it favorably to capture market share in the evolving AI voice landscape, making it a key player in the industry [30]

有网友们开始满怀悲伤地怀念与 GPT-4o 的互动体验：有人用它来排解深夜的孤独，有人用它模拟朋友间的闲聊，有人将其视为「情感寄托的港湾」。 GPT-5 的发布本应是一件好事，然而，ChatGPT 的用户却在 48 小时内，掀起了一场全球性的「反抗运动」。 8 月 8 日，OpenAI 的新模型 GPT-5 上线的同时，OpenAI 下架了 GPT-4o 等其他所有模型，强制全球用户只能使用 GPT-5 这个新模型。根据 OpenAI 掌门 Sam Altman 的介绍，GPT-5 更聪明，完成了「智商飞跃」，从大学生水准，升级到「博士级」的能力，生产力更强了。 然而，用户表示，新模型牺牲了 GPT-4o 在共情能力和提供情绪价值上的优势，回答问题时变得「冷漠」了。好像失去了一位「密友」、「恋人」。 怀念 4o 的帖子引发了广泛共鸣｜图片来源：小红书 悲伤过后，紧接着，社交媒体上，一场以「拯救 4o」（#Keep4o、#Save4o）为主题的网络声援运动迅速蔓延。 从 X(原 Twitter）到 Reddit，到小红书等等，有用户到 OpenAI 的官网账号下留言，有人制作表情包凸显 4o 的优势，有人在各平 ...

要说有什么瓜，能从去年吃到今年，那必须得有一大批传统媒体"开撕"AI大厂的一席之地。 为了防止有人还不知道这事儿，我先简单交代下背景。 这次事件的主角是一家在AI圈内声名鹊起、估值已达180亿美元的初创公司——Perplexity AI。 而站在他们对面的，则是一个由全球顶级媒体组成的"复仇者联盟"：日本最大的报业集团《读卖新闻》、英国广播公司（BBC）、新闻集团旗下的道琼斯 （《华尔街日报》母公司）和《纽约邮报》，以及早已向其发出"最后通牒"的《纽约时报》。 这些传统媒体们给Perplexity AI安的罪名是未经许可使用版权材料，他们还在自家网站上发了篇文章，专门报道了这事儿。 （图源：读卖新闻） 乍看之下，这像是一场传统媒体巨头对新兴技术公司的围剿，一场"旧时代"对"新物种"的宣战。 但当你深入了解Perplexity的工作方式后，你会发现，这场争斗远比想象中复杂。它不仅仅是关于几篇文章的版权归属，更是关于互联网信息流转方式、商 业模式乃至整个内容生态未来走向的一场深刻博弈。 （图源：雷科技制作） 这一幕，与2014年搜狐新闻们联合起来控诉今日头条何其相似。当年，作为新物种的今日头条用"千人千面"的推荐 ...

在万众瞩目中，美国人工智能巨头OpenAI终于揭开了其新一代旗舰模型GPT-5的神秘面纱。然而，这场期待已久的发布并未如预期般引发轰动，反而以一 种近乎平淡的方式落幕，成为当前人工智能领域"群雄逐鹿"格局的缩影。 发布会前夕，OpenAI首席执行官山姆·奥特曼的一番自嘲——"在GPT-5面前，我感觉自己一无是处"——虽看似谦逊，实则无形中拉高了公众的期待值，结 果自然引发了不小的嘲讽浪潮。 山姆·奥特曼的这番言论背后，是OpenAI面对的内忧外患。GPT-5的难产、核心人才的流失，以及高昂的运营成本，都让这家曾经的人工智能领头羊倍感 压力。尤其是其现金流问题，每分钟高达1.5万美元的消耗，让OpenAI的处境雪上加霜。 事实上，当前人工智能领域最具商业价值的成果，或许就是"市场营销"所打造的"AI承诺"。无论是GPT-5、Grok 4、Claude Opus 4，还是Gemini 2.5 Pro， 这些顶级模型更多地反映了行业飞速迭代、降本增效的趋势，而非颠覆性的创新。 GPT-5的发布，也暴露了OpenAI在技术瓶颈上的挣扎。高质量网络数据的枯竭、训练技巧在更大规模上的失效，以及推理模型面临的"天才诅咒"， ...

来源：钛媒体 On the HealthBench medical benchmark, Baichuan-M2 scored 60.1, surpassing OpenAI's gpt-oss120b (57.6) and other open-source leaders like Qwen3-235B and Kimi K2. In a harder subset of the benchmark, which OpenAI debuted with GPT-5 earlier this year, Baichuan-M2 scored 34.7 — making it the second model in the world to exceed the 32-point threshold. "Building models for life and creating doctors for humanity is Baichuan's mission," Wang said, calling medical foundation models the "crown jewel" of AI res ...