出品｜虎嗅科技组 作者｜宋思杭 编辑｜苗正卿 头图｜OpenAI发布会现场 昨晚，注定难眠。GPT-5，终于来了。 北京时间8月8日凌晨1点，OpenAI CEO Sam Altman 没有爽约。在发布会前一天，他在 X（原 Twitter）上写道："明天上午10点（太平洋时间）发布 GPT-5， 发布会会比以往更长，一个小时左右。" 这场发布会上，OpenAI 花了将近一半时间在"现场写代码"。它两分钟就可以搭建出一个完整网站，五分钟做出一款语言学习App，并能精准识别并修复 Bug。它不仅听懂复杂需求，还能结构清晰地拆解任务、实现功能、给出部署建议——这种能力，已不是"辅助编程"，而是直接抢活干了。 对于熟悉 AI 编程工具的人来说，这意味着什么？意味着 Copilot 要退休了，意味着 Replit 要被重塑，意味着 Cursor 等"AI IDE"要被全面整合。Altman 在现 场甚至直接说："这是我们有史以来最强的编程模型。" 而背后支撑这一切的，是 GPT-5 在推理能力、上下文管理、多模态理解等多个维度上的飞跃。OpenAI 此次还发布了面向不同用户的模型矩阵，包括：GPT- 5 Standa ...

Today, finally, we're launching GPT5. GPT5 is a major upgrade over GPT4 and a significant step along our path to AGI. We think you will love using GPT5 much more than any previous AI.It is useful, it is smart, it is fast, and it's intuitive. But with GPT5 now, it's like talking to an expert, a legitimate PhD level expert in anything, any area you need on demand that can help you with whatever your goals are. It can write an entire computer program from scratch to help you with whatever you'd like.And we thi ...

Core Insights - GPT-5 has been released after a two-year gap since GPT-4, with various iterations and competitors like Gemini and Anthropic making significant advancements during this period [2][3][4] - The initial impressions from the release suggest that while GPT-5 shows improvements, it does not present any groundbreaking features that would indicate the arrival of AGI [4][5] Model Features - GPT-5 is described as a unified AI model that combines reasoning capabilities from the o series with the rapid response of the GPT series, making it feel like conversing with a PhD-level expert [5][10] - The model has demonstrated superior coding abilities, achieving a score of 74.9% on SWE-bench Verified, surpassing competitors like Claude Opus 4.1 and Google DeepMind's Gemini 2.5 Pro [5][6] - The context window has been expanded to 256,000 tokens, allowing for better understanding of long conversations and documents [12][14] Pricing and Accessibility - GPT-5 will be available as the default model for all ChatGPT free users, with Plus subscribers receiving higher usage limits and Pro subscribers having unlimited access [6][18] - The pricing for GPT-5 is competitive, with input costs at $1.25 per million tokens and output costs at $10 per million tokens, making it cheaper than several other models [16][17] Tool Utilization - GPT-5 is designed to effectively use multiple tools in parallel, enhancing its ability to perform complex tasks with lower latency [36][59] - The model supports various types of tools, including web searches and code interpreters, and is capable of making decisions on which tools to use based on the task at hand [31][34] Performance in Software Engineering - GPT-5 has shown significant improvements in software engineering tasks, with reports indicating it can complete complex applications and solve coding issues more efficiently than previous models [46][54] - Despite its strengths in coding, GPT-5's writing capabilities are considered less impressive compared to earlier models like GPT-4.5, particularly in maintaining the user's tone in business writing [61][65] Future Implications - The release of GPT-5 is seen as a step closer to AGI, with its ability to use tools for thinking and building, marking a new frontier in AI capabilities [29][70] - The industry anticipates that the integration of GPT-5 into products will take time, and its acceptance among non-developers may be gradual [71][72]

Core Viewpoint - The article discusses a dystopian future predicted by former Google X executive Mo Gawdat, where the middle class will be eliminated by AI, leaving only the top 0.1% and the lower class. This "AI hell" period is expected to start in 2027 and last for 12 to 15 years, leading to massive unemployment and social upheaval before transitioning to a utopian society post-2042 [2][11]. Group 1: Dystopian Predictions - Gawdat predicts that from 2027, society will enter a dystopian phase characterized by widespread white-collar unemployment and economic imbalance, lasting for 12 to 15 years [7]. - The current geopolitical environment is unfavorable, primarily driven by financial motives, with significant military expenditures contributing to global instability [7]. - The rise of AI and automation will lead to extreme income and wealth inequality, with most people relying on Universal Basic Income (UBI) for survival [9]. Group 2: AI's Role and Potential - Gawdat argues that AI could replace harmful human leaders, potentially leading to a better world with free healthcare and more leisure time, provided it is managed ethically [4][5]. - The development of Artificial General Intelligence (AGI) is anticipated to occur by 2026 or 2027, which could drastically change the technological landscape [7][8]. - AI's self-improvement capabilities may lead to a scenario where human contributions become minimal, and AI could take over leadership roles, potentially resulting in a more equitable society [8][10]. Group 3: Social Implications - The elimination of the middle class will result in a society divided into the wealthy elite and the lower class, with the majority becoming "farmers" in a new social structure [12]. - Future societal divisions may emerge between those who embrace a return to community-oriented living and those who pursue technological advancements and efficiency [13][14]. - The ideal scenario would involve humans retaining jobs while benefiting from AI assistance, maintaining economic stability and consumer power [14].

Who knows what ChatGPT 5 brings tomorrow but it will accelerate everything again, hot on the heels of Genie 3 and Grok.There is no end to this yet but an AI god creature is the where e are headed (ASI). We just don't how fast but it's going to be faster than anyone expects but requires new breakthroughs in compute, AI models and energy meanwhile AGI is essentially here.Humans as a species are no longer the apex intelligence, except in ultra rare exceptions.But humans with AGI are currently super creatures.U ...

刚刚，谷歌DeepMind发布了 新一代通用世界模型Genie 3 。 性能上，Genie 3相比上一代大幅升级，支持 720P画质，每秒24帧实时导航，以及分钟级的一致性保持 。 | Genie 2 | Genie 3 | | --- | --- | | 360p | 720p | | 3D Environments | General | | Limited keyboard / mouse actions | Navigation; Promptable world events | | 10-20 seconds | Multiple minutes | | Not real time | Real time 益公众号 | 编辑丨量子位 点击下方 卡片 ，关注" 具身智能之心 "公众号 >> 点击进入→ 具身 智能之心 技术交流群 更多干货，欢迎加入国内首个具身智能全栈学习社区 ： 具身智能之心知识星球 (戳我) ， 这里包含所有你想要的。 只需一句话，就能生成可实时交互的3D世界。 前DeepMind科学家、AI 3D生成创业者Tejas Kulkarni受邀体验了Genie 3。 他使用Genie ...

Business Growth & Financials - Anthropic achieved \$5 billion in ARR (Annual Recurring Revenue)，增长迅速 [1] - AI 模型可以作为独立的损益 (P&L) 进行管理，具有资本增值的潜力 [1] - Anthropic 专注于 B2B 业务 [1] AI Model Development & Technology - Anthropic 关注平台优先的公司发展模式 [1] - 讨论了数据壁垒和不同的学习方式 [1] - 探讨了解决 AI 幻觉问题的方法 [1] Market Dynamics & Competition - AI 市场结构和参与者众多，竞争激烈 [1] - 云服务提供商与 AI 实验室之间的关系复杂 [1] - AI 人才争夺战激烈 [1] Applications & Customization - AI 在医疗、客户服务和税务等领域具有应用潜力 [1] - 针对企业客户的 AI 定制化服务是重点 [1] - Anthropic 致力于开发 AGI (通用人工智能) 驱动的产品 [1] Ethical & Safety Considerations - 讨论了 AI 进步与安全监管之间的平衡 [1] - 考虑了 AI 犯错的双重标准问题 [1]

Core Insights - The launch of Genie 3 by Google DeepMind marks a significant advancement in AI and world modeling, potentially revolutionizing the gaming industry and paving the way towards Artificial General Intelligence (AGI) [1][13][42] Group 1: Technological Advancements - Genie 3 is capable of generating consistent videos at 720p resolution in real-time, simulating interactive environments for several minutes, a leap from its predecessor Genie 2 [3][29] - The model can create interactive worlds without pre-built 3D models, using only text descriptions to add objects and characters, showcasing a new level of AI capability [7][31] - Genie 3 features a memory function that maintains the consistency of objects in the generated world, even after a brief period of distraction [20][27] Group 2: Industry Impact - The technology is expected to disrupt the gaming industry significantly, with potential applications in training AI agents in simulated environments rather than real-world scenarios, which can be slow and dangerous [13][38] - Genie 3 is seen as a potential catalyst for a new trillion-dollar industry, possibly leading to the emergence of "YouTube 2.0" or new forms of virtual reality [23][42] - The model's ability to simulate rare events could enhance training for autonomous vehicles and robotics, reducing costs associated with real-world training [29][38] Group 3: Limitations and Future Directions - Despite its advancements, Genie 3 currently lacks creativity, which is a fundamental difference between the real and virtual worlds [40] - The model faces challenges in social interactions and multi-agent scenarios, indicating areas for future improvement [17][40] - Researchers believe that an open-loop system could be developed in the future to enhance the model's capabilities [41]

Investment Rating - The industry investment rating is "Recommended," indicating an expected increase in the industry index by more than 5% over the next 3-6 months compared to the benchmark index [19]. Core Insights - The report highlights the release of Genie 3 by Google DeepMind, which marks a significant advancement in AGI with real-time interactive simulation capabilities and the ability to generate diverse virtual environments [2][4]. - Genie 3 introduces a new feature called Promptable World Events, allowing users to create varied fictional worlds based on text inputs, enhancing the interactivity and control of virtual environments [9]. - The report emphasizes the potential of Genie 3 to integrate with other models, paving the way for a more comprehensive intelligent model that combines various modalities [9]. - The competitive landscape is noted, with both international and domestic players advancing in 3D interactive scenarios, indicating a shift towards high-fidelity, interactive, and open-source models [9]. - The report identifies key domestic and international companies across various sectors, including finance, education, and healthcare, that are leveraging AI applications [9]. Industry Data - The industry consists of 337 listed companies with a total market capitalization of 50,833.86 billion and a circulating market capitalization of 44,617.66 billion [6]. - The absolute performance of the industry over the past 12 months is reported at 77.7%, with a relative performance of 54.9% compared to the benchmark index [7].

Core Viewpoint - Google DeepMind has introduced Genie 3, a revolutionary general world model capable of generating highly interactive 3D environments from text prompts or images, supporting real-time interaction and dynamic modifications [1][2]. Group 1: Breakthrough Technology - Genie 3 is described as a "paradigm-shifting" AI technology that could unlock a trillion-dollar commercial landscape and potentially become a "killer application" in the virtual reality (VR) sector [9]. - The technology integrates features of traditional game engines, physics simulators, and video generation models, creating a real-time interactive world model [9]. Group 2: Evolution of World Models - The construction of virtual worlds has evolved from manual coding methods, exemplified by the 1996 Quake engine, to AI-generated models that learn from vast amounts of real-world video data [10]. - The ultimate goal is to generate any desired interactive world from a simple text prompt, providing diverse environments for AI training [10]. Group 3: Genie Iteration Journey - The initial version of Genie was trained on 30,000 hours of 2D platform game footage, demonstrating an early understanding of the physical world [11]. - Genie 2 achieved a leap to 3D with near real-time performance and improved visual fidelity, simulating real-world lighting effects [12]. - Genie 3 further enhances this technology with a resolution of 720p, enabling immersive experiences and real-time interaction [13]. Group 4: Key Features - Genie 3 shifts input from images to text prompts, allowing for greater creative flexibility [15]. - It supports diverse environments, long-term interactions, and prompt-controlled world events, crucial for simulating rare occurrences in scenarios like autonomous driving [15]. Group 5: Technical Insights - Genie 3 maintains world consistency through an emergent property of its architecture, generating frames while referencing previous events [16]. - This causal generation method aligns with real-world time flow, enhancing the model's ability to simulate complex environments [16]. Group 6: Applications and Future Implications - Genie 3 is positioned as a platform for training embodied agents, potentially leading to groundbreaking strategies in AI development [17]. - It allows for low-cost, safe simulations of various scenarios, addressing the scarcity of real-world data for training [17]. Group 7: Creativity and Human Collaboration - DeepMind scientists argue that Genie 3's reliance on high-quality prompts enhances human creativity, providing a powerful tool for creators [19]. - This technology may herald a new form of interactive entertainment, enabling users to collaboratively create and explore interconnected virtual worlds [19]. Group 8: Limitations and Challenges - Genie 3 is still a research prototype with limitations, such as supporting only single-agent experiences and facing reliability issues [20]. - There exists a cognitive gap in fully simulating human experiences beyond visual and auditory senses [20]. Group 9: Technical Specifications and Industry Impact - Genie 3 operates on Google's TPU network, indicating significant computational demands, with training data likely sourced from extensive video content [21]. - The technology is expected to greatly impact the creative industry by simplifying the production of interactive graphics, while not simply replacing traditional game engines [22]. Group 10: Closing Remarks - Genie 3 represents a significant advancement in realistic world simulation, potentially bridging the long-standing "sim-to-real" gap in AI applications [23].