日前，腾讯云金融数智峰会在北京召开，聚焦数智化在金融领域的前沿应用与深度价值挖掘。腾讯云天御总经 理、首席科学家李超和腾讯云天御金融风控产品专家吴恩斌分别就当前金融行业在风控领域普遍面临的数据要素 流通、模型对抗瓶颈等核心痛点，分享了腾讯云天御基于模型和数据的技术应对举措及整体智能风控防线的构 建，以期通过腾讯云天御的风控最佳实践，为行业业务反欺诈提供参考与借鉴。 针对数字化浪潮与风险升级、监管与效率的双重压力以及普惠金融的经济刚需的行业痛点，腾讯云天御提出 了"数据-模型-场景"三层架构：高质量数据是智能风控的基础支撑，模型是智能风控的核心引擎，而应用则是智 能风控的落地价值，三层递进形成智能风控闭环，全方位增强防控实力，推动金融行业稳健发展。 其中，模型能力决定风控上限。金融行业是天然的数据密集型行业，有场景、有需求、有技术沉淀，是大模型绝 佳的"练兵场"。为确保风险控制的持续有效性，金融机构需针对不同业务领域建立合适的风控模型，并根据风险 情况不断迭代模型。 在全球金融科技浪潮推动行业深刻变革的背景下，以AI大模型为代表的核心驱动力，正逐步渗透到金融服务各个 环节。创新技术的涌现赋能网络安全的攻防两端， ...

全流程效率的提升。传统金融服务中，从客户身份核验到风险评估，从客服响应到贷后管理，多依赖人 工规则与单一模型，不仅处理效率低下，且难以应对复杂场景。大模型通过自然语言处理（NLP）、计 算机视觉（CV）等多模态能力的融合，可实现快速响应。例如在贷款审批环节，在大模型支持下的智 能化、自动化系统能在较短时间内完成多维度数据核验与风险定价，大幅降低人工成本。 齐鲁晚报·齐鲁壹点记者 尚青龙 当生成式AI技术从实验室快速走向产业应用，消费金融行业正站在数智化转型的关键节点。AI大模型 以其强大的多模态理解、自主决策与持续进化能力，正在重构金融服务的底层逻辑，从风险控制到客户 服务，从精准营销到运营效率，全方位推动行业进入AI金融新时代。在这一变革浪潮中，海尔消费金 融（简称"海尔消金"）通过"AI-First"战略的深度落地，助力消金业务智能化升级，为用户提供更优质、 更体贴的金融服务。 当前，在消费金融领域，AI大模型的价值正通过三个维度深度释放。 如在智能客服领域，大模型推动服务从被动响应到主动预判。海尔消金将客服系统从传统 NLP升级至 大模型全流程模式，构建起问答对话智能体与任务对话智能体，其中语音机器人累 ...

Market Overview - The market showed mixed performance with the Shanghai Composite Index down 0.68% to 3591.26 points, the Shenzhen Component Index down 0.45% to 11152.77 points, and the ChiNext Index up 0.43% to 2377.84 points [1][2]. Sector Performance - Strong sectors included assisted reproduction, AI technology, and liquid cooling servers, while steel, oil and gas, and coal mining sectors declined [4]. Capital Flow - Major capital inflows were observed in the computer, communication, and pharmaceutical sectors, while outflows were noted in non-ferrous metals, food and beverage, and public utilities [5]. Stock Specifics - Notable net inflows were seen in stocks such as Cambridge Technology (14.63 billion), Easy Point (12.33 billion), and Changying Precision (11.54 billion) [6]. - Conversely, significant net outflows were recorded for Kweichow Moutai (12.81 billion), CATL (8.85 billion), and Midea Group (7.26 billion) [7]. Institutional Insights - China Galaxy Securities indicated that the real estate sector may see overall valuation recovery, with leading companies benefiting from lower financing costs and high market share in core areas [9]. - CITIC Securities highlighted a shift in policy supporting the innovative drug and medical device industries, suggesting potential valuation and performance recovery in the medical device sector [9]. - Huatai Securities provided insights on the AI large model industry chain, noting a new growth phase driven by token growth, increasing server computing demand, and a competitive landscape fostering industry advancement [10].

Group 1 - Midea plans to invest 50 billion yuan in cutting-edge technology over the next three years, focusing on areas such as AI models, new energy, robotics, and embodied intelligence [1][2] - The investment strategy is a response to increasing competition in the home appliance industry, particularly from brands like Xiaomi and emerging players [2][3] - Midea's chairman, Fang Hongbo, emphasizes the need to shift from a home appliance focus to a technology-driven approach, indicating a strategic pivot towards B2B operations [4][10] Group 2 - Midea's R&D expenditure has been significant, with nearly 60 billion yuan invested over the past five years, and 16.2 billion yuan allocated in 2024 alone, reflecting a commitment to technological advancement [12][13] - The company has a history of acquisitions to strengthen its supply chain and technological capabilities, including the complete acquisition of KUKA, a leading robotics company [14][16] - Despite these efforts, Midea's B2B revenue remains only 25.67% of total revenue, indicating challenges in achieving its strategic goals [22] Group 3 - The competitive landscape is intensifying, with domestic rivals like Ecovacs and Roborock entering the robotics sector, which Midea has historically dominated [27][29] - Midea's competitive advantages stem from its manufacturing efficiency and technological investments, but the rise of new entrants poses a threat to its market position [28][29] - The company faces pressure to innovate and maintain its leadership in the face of rapidly evolving competition in the robotics and home appliance markets [27][29]

Core Insights - The venture capital industry is showing signs of recovery, with significant improvements in fundraising and investment activities compared to previous years [1][2] - The industry is undergoing profound changes, with a focus on nurturing high-quality projects and adapting to new policies that encourage long-term investments [1][2] Fundraising - In the first half of this year, the amount pledged by institutional LPs increased by 50% year-on-year, ending a continuous decline since 2020 [1] - The overall scale of investment and financing saw a reduction in decline from 50% in the same period last year to just 5% [1] Investment - A total of 131 Chinese companies went public in A-shares, Hong Kong, and US markets in the first half of 2025, with 56% receiving support from VC/PE [1] - The Hong Kong IPO market has shown strong performance, and diverse exit strategies such as mergers and acquisitions are becoming more mature [1] Industry Transformation - The emergence of high-quality projects like DeepSeek, Yushu Technology, and Yingshi Innovation highlights the international competitiveness of Chinese technology brands [1] - Policies such as "Venture Capital Seventeen Articles" and "State Council No. 1 Document" are being introduced to attract long-term capital from banks, insurance funds, and social security [1] Value Creation - The venture capital industry is encouraged to adopt a long-term perspective, moving beyond short-term financial models to discover and nurture value [2][3] - There is a call for venture capital firms to extend their value beyond mere funding, acting as long-term partners to startups and enhancing post-investment support [2] Ecosystem Development - The industry is urged to foster an open and inclusive innovation ecosystem, collaborating with research institutions, industry resources, and local governments to support the growth of innovative technologies [3] - The focus is on creating a vibrant and resilient technological innovation ecosystem that supports the entire lifecycle of innovative enterprises [3]

Core Insights - The emergence of AI large models like DeepSeek has sparked interest in their potential applications across various industries, although a universal application has yet to be identified [1] - Experts suggest that focusing on B2B applications of large models may be a more viable approach in the short term [1] - The 2025 World Artificial Intelligence Conference showcased various AI technologies, including humanoid robots and digital humans, highlighting the advancements in AI [1] Group 1: AI Large Models and Digital Transformation - AI large models serve as a new digital foundation and intelligent application engine, enhancing operational efficiency and decision-making capabilities in enterprises [3] - Traditional digitalization focuses on speeding up processes, while AI large models enable clearer problem identification and more accurate decision-making [3] - The integration of AI large models into business processes is essential for achieving intelligent operations, moving beyond mere data analysis [3] Group 2: Enterprise Applications of AI - Inspur's Haiyue large model emphasizes deep integration with business scenarios, offering enterprise-level applications rather than standalone AI tools [3][5] - The Haiyue framework combines models, scenarios, processes, knowledge, and permissions to create a comprehensive enterprise agent system [5] - The focus is on embedding AI capabilities into business workflows, enhancing efficiency, risk control, and value creation [5] Group 3: Future of AI in Business - Companies need a hybrid AI capability that combines discriminative and generative AI, with high-quality data being crucial for maximizing AI application potential [6] - The shift from a competition of model capabilities to a focus on deepening scene value will determine the long-term success of AI in industry [6] - The essence of this transformation is to align technology with business needs, fostering a digital ecosystem that supports intelligent operations [6]

Group 1 - The AI glasses were a major highlight at the World Artificial Intelligence Conference (WAIC), attracting significant attention and exceeding the booth's capacity for audience engagement [1] - The AI glasses demonstrated impressive smart interaction capabilities, accurately recognizing voice commands for tasks such as prompting and payment, and featuring high-quality photo and video functions [1] - Real-time translation capabilities of the AI glasses were noteworthy, effectively converting foreign languages into Chinese and displaying them on the lenses, facilitating cross-border communication [1] Group 2 - Major companies, including Alibaba, are entering the AI glasses market, with Alibaba's self-developed Quark AI glasses expected to launch within the year, integrating with its ecosystem for various functionalities [2] - The AI glasses market is still in its early stages, with no single product achieving sales of over one million units, except for Meta, which has subsidies influencing its sales [2] - The development of smart glasses is entering a rapid growth phase, presenting both opportunities and challenges, as noted by industry leaders [2]

Core Insights - The article discusses the release of the new AI model Qwen3-30B-A3B-Instruct-2507, which showcases significant improvements in performance and efficiency compared to its predecessor and other industry models [1][2]. Performance Improvements - The new model operates in a non-thinking mode, activating only 3 billion parameters while achieving performance comparable to leading closed-source models like Google's Gemini 2.5-Flash and OpenAI's GPT-4o [2][4]. - Performance metrics show substantial improvements, with AIME25 scores increasing from 21.6 to 61.3 and Arena-Hard v2 scores rising from 24.8 to 69.0 [3][4]. Benchmark Comparisons - In benchmark tests, the Qwen3-30B-A3B-Instruct-2507 model performs on par or surpasses models such as DeepSeek-V3-0324 across various categories [4][10]. - The model's average performance in knowledge benchmarks is 62.8, with specific scores like MMLU-Pro at 78.4 and GPQA at 70.4 [10]. Enhanced Capabilities - The model exhibits significant advancements in general capabilities, including instruction following, logical reasoning, text comprehension, mathematics, science, programming, and tool usage [13][27]. - It has improved multilingual knowledge coverage and can generate higher quality text aligned with user preferences [13][27]. Open Source and Accessibility - The model has been made open-source and is available on platforms like HuggingFace and QwenChat, allowing broader access and community support [16][17]. - Users have reported successful implementations of the model on consumer-grade GPUs, such as the RTX 3090, highlighting its accessibility [24][23]. Contextual Understanding - The model's long-context understanding has been enhanced, now supporting up to 256K tokens, which is a significant improvement for complex tasks [28][27]. Industry Impact - The rapid advancements in model efficiency and performance are noted as a significant trend in the AI industry, with the Qwen team setting a high bar for future developments [7][35].

Group 1 - The core viewpoint is that the innovative drug-themed ETFs have shown remarkable performance, with six ETFs increasing over 100% this year, all belonging to the innovative drug sector [1][4]. - The top-performing ETFs on July 29 include various innovative drug-themed ETFs, with the Hong Kong Stock Connect Innovative Drug ETF rising by 5.06%, marking it as the largest gainer of the day [4][6]. - The innovative drug sector is expected to experience a "Davis Double Play" due to supportive policies, international breakthroughs, improved R&D efficiency, and continuous capital inflow, indicating a shift from defensive to offensive investment strategies [4][5]. Group 2 - On July 29, ten ETFs had transaction volumes exceeding 10 billion yuan, with the Hong Kong Stock Connect Innovative Drug ETF and the Hong Kong Securities ETF being particularly active [3][9]. - The Hong Kong Innovative Drug ETF has seen significant growth, with its scale reaching 10.31 billion yuan and a year-to-date increase of over 750% [5]. - The bond ETFs have also attracted substantial inflows, with the South China Science and Technology Bond ETF leading with a net inflow of 25.56 billion yuan on the previous trading day [11][12]. Group 3 - The largest drop in ETFs on July 29 was observed in the 180 Governance ETF, which fell by 2.71%, alongside several other cross-border and bank-themed ETFs [2][8]. - The trading activity indicates a strong market liquidity, with the average daily transaction volume for the Hong Kong Innovative Drug ETF exceeding 11 billion yuan in July [5]. - The overall market sentiment remains positive, with expectations for indices to continue reaching new highs in the second half of the year, supported by domestic policies and liquidity [13].

Group 1 - The core viewpoint of the article is that Jin Yi Culture (002721) is actively engaging in digital RMB products and AI-related projects through its subsidiary, Kaike Weishi [1] - Kaike Weishi has partnered with several banks to develop digital RMB products, indicating a strategic move towards digital finance [1] - The company has won bids for AI large model projects and is developing software that integrates AI technology for applications in smart marketing, wealth management, intelligent investment advisory, and customer service [1] - Despite these advancements, the overall scale of these initiatives is currently small [1]