Core Insights - GPT-5 was launched after multiple delays, showcasing improvements in intelligence, programming, and task reasoning, but did not significantly surpass its predecessor GPT-4 in performance [1][2] - The pricing strategy for GPT-5 is a major highlight, with API call costs being significantly lower than competitors, indicating strong cost-effectiveness in the current large model market [1][6] - The release of GPT-5 reflects a shift in focus from technical breakthroughs to practical applications of AI in daily life, as the excitement around large model parameters diminishes [2][3] Group 1: Evolution of GPT Models - The GPT series has evolved from GPT-1 in 2018 to GPT-5, with significant milestones including the jump from 1.5 billion parameters in GPT-1 to 175 billion in GPT-3, and the introduction of multimodal capabilities in GPT-4 [3][4] - The narrative around large model parameters is shifting towards a focus on multimodal capabilities and practical applications in sectors like healthcare and education [3][4] Group 2: Market Dynamics and Competition - The launch of GPT-5 coincides with a wave of new model releases from competitors, indicating a competitive landscape where companies are racing to capture user attention [7][8] - The slowdown in significant upgrades for large models suggests a maturation of the market, with recent models showing limited improvements compared to earlier iterations [7][8] Group 3: AI Agent Emergence - The rise of AI Agents, which are specialized applications built on large models, is shifting the commercialization focus from general-purpose models to tailored solutions for specific industries [11][12] - OpenAI anticipates that revenue from AI Agents will surpass that from ChatGPT by the end of 2025, reflecting a significant shift in business strategy [11][12] Group 4: OpenAI's Strategic Moves - OpenAI has introduced multiple versions of GPT-5, including a mini and a pro version, indicating a move towards a more customizable and tiered pricing model [6][13] - The company has also resumed releasing open-source models, suggesting a response to industry demands for transparency and collaboration [13]

Core Insights - MiroMind ODR (Open Deep Research) is introduced as a powerful open-source deep research model, achieving a GAIA score of 82.4, surpassing other models like OpenAI's Deep Research and Manus [1][4][32] - The project is fully open-source, including its core model, data, training processes, AI infrastructure, and DR agent framework, allowing for reproducibility [3][4][15] - The team plans to maintain a monthly update schedule for open-source contributions, indicating a commitment to continuous improvement and community engagement [4][21] Performance Metrics - MiroMind ODR achieved a GAIA score of 82.4, while OpenAI's Deep Research scored 67.4 and Manus scored 73.3, highlighting MiroMind's superior performance [4][19] - The model MiroThinker, part of the ODR project, has shown state-of-the-art performance with a score of 60.2% on GAIA-Text-103 [19][21] Project Components - MiroMind ODR consists of four sub-projects: MiroFlow (Agent framework), MiroThinker (model), MiroVerse (data), and MiroTrain (training infrastructure), each contributing to the overall functionality and performance of the deep research model [15][21] - MiroFlow supports multiple mainstream tool calls and extends large language models for tool-assisted deep research reasoning [18][21] - MiroVerse provides 147,000 open-source training datasets, focusing on community feedback and continuous updates [21][32] Leadership and Vision - Dai Jifeng, a prominent figure in the project, has a strong academic background and extensive experience in computer vision and deep learning, having published over 80 papers with significant citations [26][30] - The mission of MiroMind is to develop self-aware digital entities that evolve with the community, aiming for safe and beneficial AGI [30][32]

RT David Sacks (@DavidSacks)A BEST CASE SCENARIO FOR AI?The Doomer narratives were wrong. Predicated on a “rapid take-off” to AGI, they predicted that the leading AI model would use its intelligence to self-improve, leaving others in the dust, and quickly achieving a godlike superintelligence. Instead, we are seeing the opposite:— the leading models are clustering around similar performance benchmarks;— model companies continue to leapfrog each other with their latest versions (which shouldn’t be possible i ...

What do you think about Meta buying up talent with huge salaries and and the heating up of this battle for talent. You know, there's a strategy that that Meta is taking right now. I think that from my perspective at least, I think the people that are real believers in the mission of AGI and what it can do and understand the real consequences both good and bad from that and what's what that responsibility entails.I think they're mostly doing it to be like myself to be on the frontier of that research. So, yo ...

RT Steve Jurvetson (@FutureJurvetson)Updated with the latest GPT-5Grok is still in a league of its own.And learning more rapidly.In the path to AGI, process >> product. https://t.co/yZMRkH8sbv ...

以下文章来源于新皮层NewNewThing ，作者王杰夫 新皮层NewNewThing . 关注AI，提供洞察。 2025.08. 09 本文字数：2684，阅读时长大约4分钟 作者 | 新皮层NewNewThing 王杰夫 8月8日凌晨，OpenAI终于发布了新一代有整数编号的GPT模型GPT-5，距上一代GPT-4发布已经过去2年4个月零24天。 过去，每一代GPT模型都标志着某种技术上的突破：随着参数规模扩大，GPT-3「涌现」出了GPT-2没有的智能水平；到了GPT-4，模型开始具有 图像理解相关的多模态能力。相比之下，打磨了2年之久的GPT-5却显得有些「平庸」：OpenAI说GPT-5是个「博士」，但在各项能力上，除了更低 的幻觉——比GPT-4o低约45%，比OpenAI o3低约 80%，GPT-5没有展示出先前模型没有的能力，AGI也没有到来。 幻觉降低是模型最大优化。 甚至连OpenAI自己都不再将GPT-5称作「模型」，而是将其定义为「一个统一的系统」（One unified system）。 不过，对于这个曾被寄予AGI理想的「模型」，OpenAI给出了所有竞争对手中最低的调用价格：G ...

Core Viewpoint - The article discusses the mixed reactions to the release of GPT-5, highlighting both excitement and criticism from users, indicating high expectations for AGI [4][5][6]. Group 1: GPT-5 Release and User Reactions - The release of GPT-5 has sparked widespread discussion, with some users expressing disappointment while others praise its capabilities [4][5]. - The anticipation for GPT-5 has evolved into expectations for AGI within a short span since the launch of ChatGPT in November 2022 [6]. Group 2: OpenAI's Response and Future Plans - OpenAI acknowledges the challenges faced during the rollout of GPT-5 and commits to improving system stability and user experience [9][10]. - OpenAI plans to double the usage limits for ChatGPT Plus users and will allow them to continue using GPT-4o based on usage patterns [14][15]. - The company is working on enhancing the user interface to allow easier model switching and to clarify which model is responding to queries [14][15]. Group 3: Model Features and Improvements - GPT-5 will automatically enable reasoning capabilities, and future updates will improve the switching process between models [15][32]. - A new voice model has been introduced, enhancing instruction adherence and response speed [23]. - OpenAI is focusing on better handling of biases and aims to make GPT-5 mini feel more personable [29]. Group 4: User Feedback and Customization - OpenAI recognizes the diverse preferences among users for different model versions and is considering options for simultaneous access to multiple models [16][17]. - The company is exploring the possibility of unlimited usage for reasoning features for Plus users [35][36].

Core Viewpoint - MiroMind ODR (Open Deep Research) is introduced as a powerful open-source deep research model, achieving a GAIA test score of 82.4, surpassing other models like OpenAI's Deep Research and Manus [2][5]. Group 1: Model Performance and Features - MiroMind ODR has the highest performance score of 82.4 in GAIA validation, outperforming models such as OpenAI Deep Research (67.4) and Manus (73.3) [2][5]. - The model is fully open-source and reproducible, with all core components, data, training processes, and frameworks available for public access [4][5]. - The project team plans to maintain a monthly update schedule for open-source contributions, indicating ongoing development and improvement [5]. Group 2: Sub-Projects Overview - MiroMind ODR consists of four sub-projects: MiroFlow (Agent Framework), MiroThinker (Model), MiroVerse (Data), and MiroTrain (Training Infrastructure) [20]. - MiroFlow supports multiple mainstream tool calls and extends large language models, achieving stable reproducibility with a performance score of 82.4 on GAIA [22]. - MiroThinker is a large language model that natively supports tool-assisted reasoning, demonstrating top performance in GAIA [23]. - MiroVerse provides 147,000 open-source training datasets, focusing on community feedback and continuous updates [26]. - MiroTrain supports stable and efficient training for deep research models, covering the entire training process [27]. Group 3: Development Team and Leadership - Dai Jifeng, a prominent figure in the project, has a strong academic background and extensive experience in computer vision and deep learning, with over 80 published papers and more than 60,000 citations [32][36]. - His previous roles include positions at Microsoft Research Asia and SenseTime, and he has returned to academia as an associate professor at Tsinghua University [40][41]. - The project aims to contribute to AGI (Artificial General Intelligence) research, with a mission to create self-aware digital entities that evolve with the community [45][47].

Group 1 - The core viewpoint is that the AI industry is at a critical juncture, with the potential for the emergence of "super applications" akin to TikTok, driven by intelligent agents [2][8] - The landscape of large models is consolidating, with a few major players dominating, while new startups are emerging in the AI space [3][4] - The demand for computing power remains high, particularly for inference chips, which are crucial for the operation of intelligent agents and AI applications [4][5] Group 2 - The Chinese chip market is expected to undergo consolidation, leading to significant merger and acquisition opportunities, with some AI chip startups likely to go public [5][6] - The focus on intelligent agents is seen as a major investment opportunity, with the potential for hundreds of unicorns in China and thousands globally [8][10] - The evaluation of AI projects emphasizes the importance of user willingness to pay and the product's ability to deliver tangible results, distinguishing AI products from traditional SaaS tools [13][14]

Group 1 - Microsoft CEO Satya Nadella announced the launch of GPT-5 across multiple platforms, including Microsoft 365 Copilot, GitHub Copilot, and Azure AI Foundry, highlighting significant breakthroughs in reasoning, coding, and chatting capabilities [2] - Elon Musk warned that OpenAI could potentially "swallow" Microsoft, indicating a competitive landscape in AI development [2] - xAI's co-founder Yuhuai Wu claimed that despite a smaller team, they are leading in many aspects, with Grok4 being the first unified model globally and outperforming GPT-5 in benchmark tests [2][2] - Musk expressed support for xAI's progress and mentioned that Grok5 is expected to launch by the end of the year [2] Group 2 - Reports indicated that Tesla is disbanding its Dojo supercomputer team and will rely on external technology partners like NVIDIA, AMD, and Samsung, with Musk stating that focusing resources on different AI chip designs is not practical [2] - Research from SignalFire revealed that Anthropic's engineering team is expanding at a rate significantly higher than competitors, with a hiring-to-loss ratio of 2.68, compared to OpenAI's 2.18, Meta's 2.07, and Google's 1.17 [2]