理想MindGPT 3.1被大大低估核心3个锚点： 1.理想卡片大师这种Agent，底层能力来自MindGPT 3.1。卡片大师背后是理想的AI能力，产品化能力。卡片大师有很大用户价值潜力（大多数人不以为 然）。 2.MindGPT 3.1的ASPO借鉴了DeepSeek R1 GRPO选择性学习优于全局学习的根本思想，并进行了创新优化。这再次印证着理想有充分能力快速学习AI社 会最优秀的核心思想，再内化到自己能力进行原创。 3.在以上两点的前提下，除了个别号转发了理想自己的技术解读通稿（详见:《 理想同学MindGPT 3.1发布：极速推理的智能体语言模型 》），TOP2信息 茧房内，没有任何号进行长文解读，看好理想人士圈几乎没有讨论度，仅微博用户肉鸡Frank进行了高度正面评价。 以下为细化论述： MindGPT 3.0是一个有思维链的深度思考模型，MindGPT 3.1是一个推理速度很快的智能体语言模型。 MindGPT 3.1最高速度可达每秒200 tokens, 较MindGPT 3.0提高近5倍，作为对比GPT-4o-204-05-13每秒最大值为79.87 tokens。 | Statistic ...

新京报贝壳财经讯(记者罗亦丹)8月21日，DeepSeek在官方公号发文称，正式发布DeepSeek-V3.1。本次 升级包含以下主要变化：一个模型同时支持思考模式与非思考模式的混合推理架构；相比DeepSeek-R1- 0528，DeepSeek-V3.1-Think能在更短时间内给出答案；通过Post-Training优化，新模型在工具使用与智 能体任务中的表现有较大提升，因此拥有更强的Agent能力。 目前，官方App与网页端模型已同步升级为DeepSeek-V3.1。用户可以通过"深度思考"按钮，实现思考模 式与非思考模式的自由切换。DeepSeek在官方发文称，DeepSeek-V3.1的发布是"迈向Agent时代的第一 步"。 ...

Core Insights - DeepSeek has officially launched DeepSeek-V3.1, marking a significant step towards the era of intelligent agents with its hybrid reasoning model and 671 billion parameters, surpassing previous models like DeepSeek-R1 and Claude 4 Opus [1][12][18] Model Performance - DeepSeek-V3.1 demonstrates faster reasoning speeds compared to DeepSeek-R1-0528 and excels in multi-step tasks and tool usage, outperforming previous benchmarks [3][6] - In various benchmark tests, DeepSeek-V3.1 achieved scores of 66.0 in SWE-bench, 54.5 in SWE-bench Multilingual, and 31.3 in Terminal-Bench, significantly surpassing its predecessors [4][15] - The model scored 29.8 in the Humanity's Last Exam, showcasing its advanced reasoning capabilities [4][16] Training and Architecture - The model utilizes a hybrid reasoning mode, allowing it to switch between reasoning and non-reasoning modes seamlessly [6][12] - DeepSeek-V3.1-Base underwent extensive pre-training with 840 billion tokens, enhancing its contextual support [6][13] - The training process involved a two-stage long context expansion strategy, increasing the training dataset significantly [13] API and Accessibility - Starting September 5, a new API pricing structure will be implemented for DeepSeek [7] - Two versions of DeepSeek-V3.1, Base and standard, are available on Hugging Face, supporting a context length of 128k [6][14] Competitive Landscape - DeepSeek-V3.1 has been positioned as a strong competitor to OpenAI's models, particularly in reasoning efficiency and coding tasks, achieving notable scores in various coding benchmarks [12][20][23] - The model's performance in coding tests, such as Aider, reached 76.3%, outperforming Claude 4 Opus and Gemini 2.5 Pro [16][19]

当谷歌的AI实习生独立编写代码、特斯拉的人形机器人Optimus开始自主规划仓库路径，一个尖锐的问题正浮出水面：人类在智能体狂潮中的角色， 究竟是被取代的"旧劳动力"，还是进化为更高维的"规则缔造者"？ 福布斯中国AI影响力人物、中国人工智能领军科学家刘志毅在其新著《智能体时代》一书揭开了这场分工革命的底层逻辑：智能体技术的崛起正在对工 作的时间和空间维度进行一场彻底的重构。信息时代延续了这一模式，尽管工作内容从体力劳动转向了脑力劳动。然而，智能体时代正在彻底打破这种 时空耦合，使工作从固定的物理空间和规定的时间段中解放出来。 信息时代延续了这一模式，尽管工作内容从体力劳动转向了脑力劳动。然而，智能体时代正在彻底打破这种时空耦合，使工作从固定的物理空间和规定 的时间段中解放出来。 2023 年，OpenAI的GPT-4 模型问世后，知识工作的本质开始发生转变。最初，人们将这些工具视为高级文本生成器，但随着垂直领域专用智能体的 崛起，它们逐渐演变为真正的"知识合作伙伴"。纵观历史，每一次重大技术革命都会重塑工作的基本构成。工业革命将农业劳动者转变为工厂工人，互 联网革命创造了大量数字工作岗位。而智能体革命则是对 ...

Core Insights - The article discusses the evolution of Chaojuvian from a survival phase to an aggressive growth phase, marked by the launch of the "Chaojuvian 2.0" plan focused on AI-driven business process reconstruction [1][2][4] - Chaojuvian aims to leverage four key technology areas: AI, data, computing power, and energy, to build a new ecosystem and drive the emergence of intelligent entities [5][7] Group 1: Company Development - Chaojuvian's business scale has increased from approximately 10 billion to over 40 billion in three years, with a customer base expanding from around 2,000 to over 24,000 [2][4] - The company has optimized its underlying capabilities through a comprehensive redesign of processes and organization, laying a foundation for future growth [4] Group 2: Technological Focus - The six key technology elements identified by Chaojuvian that will drive future changes across industries include AI, data, computing power, energy, materials, and biotechnology [5] - Chaojuvian's strategy includes a dual-ecosystem approach for computing power, integrating both Eastern and Western technological ecosystems [7] Group 3: New Product Launches - Chaojuvian has introduced multiple new products and solutions across three main areas: computing power, digital transformation, and energy [9][12] - The computing division has launched six major products, including upgraded liquid-cooled servers and AI integration platforms [12] Group 4: Energy Solutions - In the energy sector, Chaojuvian aims to create a digital and efficient thermal management system for electric vehicles, charging stations, and power grids [13] - New products in the energy field include split charging hosts and liquid-cooled ultra-fast charging terminals, along with AI-driven operational maintenance services [13] Group 5: Future Direction - Chaojuvian positions itself as an ecosystem-oriented enterprise, emphasizing exploration as a key theme for its next phase of development [16]