Welcome to the 【AI Daily】column! This is your daily guide to exploring the world of artificial intelligence. Every day, we present you with the hottest topics in the AI field, focusing on developers and helping you understand technological trends and innovative AI product applications.
New AI Products Learn More: https://top.aibase.com/
1. Zhipu AI Launches New Domain Z.ai and Open-Sources 32B/9B Series GLM Models
The Zhipu technology team recently announced the open-sourcing of its 32B and 9B series GLM models and the launch of its new interactive experience platform, Z.ai. These models are under the MIT license, supporting commercial use and increasing developer freedom. The GLM-4-32B-0414 model excels in multiple tasks, even surpassing some larger models.
【AiBase Summary:】
🆓 The open-sourced 32B and 9B series GLM models are under the MIT license, supporting commercial use and free distribution.
⚡ The GLM-Z1-32B-0414 inference model achieves a speed of 200 Tokens/second, the fastest among domestic commercial models.
🧠 The GLM-Z1-Rumination-32B-0414 model, through deep thinking and search tool integration, enhances complex task processing capabilities.
2. Alibaba Cloud's AI Open-Source Community, ModelScope, Launches MCP Plaza
Alibaba Cloud's AI open-source community, ModelScope, has launched the new MCP Plaza, becoming the largest Chinese MCP community. The platform hosts over a thousand MCP services and exclusively released new services such as Alipay and MiniMax, providing AI developers with abundant resources and driving innovation in AI applications. The Alipay MCP service focuses on intelligent payment scenarios, simplifying the AI commercialization process, while MiniMax integrates various generative models into a unified tool, expanding application scenarios.
【AiBase Summary:】
🛠️ ModelScope launches MCP Plaza, becoming the largest Chinese MCP community, providing abundant AI development resources.
💳 The Alipay MCP service focuses on AI payment scenarios, simplifying the process of bringing AI services to market.
📊 ModelScope also launched the MCPBench tool to help developers evaluate the effectiveness and resource consumption of MCP services.
Details: https://modelscope.cn/mcp
3. OpenAI Releases GPT-4.1 Series Models: Capabilities Significantly Exceed Previous Generation
On April 15th, OpenAI officially released the GPT-4.1 series models, including GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano. This series of models has made significant breakthroughs in programming capabilities, instruction understanding, and long-text processing, significantly improving development efficiency and reducing costs. GPT-4.1 set new records in multiple benchmark tests, especially excelling in programming scenarios with a 40% increase in code generation speed.
【AiBase Summary:】
💻 GPT-4.1 series models have achieved a 40% speed increase in programming capabilities, and user query costs have been reduced by 80%.
📊 In the SWE-bench Verified test, GPT-4.1 scored 54.6%, a 21.4 percentage point improvement over its predecessor.
⚖️ CoCounsel, a legal assistant integrated with GPT-4.1, has improved the accuracy of multi-document review by 17%.
4. Nuclear-Level Evolution! Keling AI Enters the 2.0 Era
The release of Keling AI 2.0 marks a significant advancement in multi-modal AI creation. This update includes a complete overhaul of two core models: Keling 2.0 (Master Edition) and Ketu 2.0. The former achieves a qualitative leap in video generation, while the latter breaks through the response capability to complex instructions in static image generation. In addition, the new multi-modal editing function allows users to more flexibly edit videos and images, improving creative efficiency and output quality.
【AiBase Summary:】
🎥 Keling 2.0 (Master Edition) achieves a qualitative leap in video generation, accurately responding to complex temporal descriptions.
🖼️ Ketu 2.0 breaks through the ability to follow complex instructions in static image generation, supporting various artistic styles.
✂️ The newly added multi-modal editing function allows users to flexibly edit videos and images, improving creative efficiency.
Details: https://app.klingai.com/cn/release-notes
5. Kimi, the Dark Side of the Moon, Launches its First Content Community, Ushering in a New Era of User Interaction
The Dark Side of the Moon recently announced that its first content community product, Kimi, is undergoing gray-scale testing, aiming to improve user experience and retention. Led by a team with a Tsinghua University background, the company launched a smart assistant supporting up to 2 million Chinese characters last year. Despite securing $1 billion in funding and a $2.5 billion valuation, its influence has waned with increasing market competition.
【AiBase Summary:】
🌟 The Dark Side of the Moon launches the Kimi content community product, undergoing gray-scale testing to enhance user experience.
📈 The company is led by a Tsinghua team and has participated in the development of several well-known large models, with its smart assistant sparking much discussion.
⚖️ The content community will face challenges regarding user content compliance and authenticity, while also needing to maintain user activity.
6. Zhipu AI Officially Initiates IPO Process, Ushering in a New Chapter for the "Big Six" Large Models
Zhipu AI has officially initiated its Initial Public Offering (IPO) process, becoming a focal point in China's large model field. As one of the "Big Six" large models, this milestone marks a significant step for Chinese AI startups entering the capital market. Leveraging Tsinghua University's technological accumulation, Zhipu AI has achieved remarkable success in large model R&D and commercial application, and is expected to occupy a more important position in the global AI market.
【AiBase Summary:】
📈 Zhipu AI completed its guidance registration with the Beijing Securities Regulatory Bureau on April 14, 2025, with CICC as its guidance institution, marking the start of its IPO process.
💡 As one of the "Big Six" large models, Zhipu AI continues to lead in technological innovation, launching multiple advanced models and attracting numerous enterprises and developer users.
💰 Since its establishment in 2019, Zhipu AI has completed over 12 rounds of financing, with a valuation exceeding 20 billion yuan, and investors including Tencent, Alibaba, and Meituan.
7. Tencent Cloud's Large Model Knowledge Engine Upgrade: Supports MCP Protocol, Empowering Application Development
On April 14th, Tencent Cloud announced a significant upgrade to its large model knowledge engine, now supporting the MCP protocol. This update allows users to utilize platform-selected MCP plugins or insert custom plugins as needed, greatly enhancing application functionality and flexibility. Tencent Cloud has selected many high-quality MCP Servers covering various application scenarios to meet user needs in information retrieval and web processing.
【AiBase Summary:】
🛠️ Tencent Cloud's large model knowledge engine supports the MCP protocol, allowing users to directly call selected or custom plugins.
🌐 Selected MCP plugins cover various application scenarios, including web deployment, geographic information, and real estate listings.
⚙️ Users can configure MCP SSE services and flexibly integrate them into applications to improve development efficiency.
8. XPeng Motors Initiates Development of 720 Billion Parameter "XPeng World Base Model"
XPeng Motors held an AI technology sharing session in Hong Kong, unveiling its ultra-large-scale autonomous driving large model with 720 billion parameters – the "XPeng World Base Model." This model uses a multi-modal architecture, aiming to improve autonomous driving technology to surpass human capabilities. XPeng Motors has also built China's first 10,000-card intelligent computing cluster to support the research and application of this technology.
【AiBase Summary:】
🚀 XPeng Motors launches a 720-billion-parameter autonomous driving large model, aiming to achieve driving technology that surpasses human capabilities.
🖥️ Through the "cloud-end model factory," XPeng has achieved a full-link process from cloud training to vehicle-end deployment.
📊 XPeng Motors has built China's first 10,000-card intelligent computing cluster, with a computing power scale of 10EFLOPS, supporting AI infrastructure construction.
9. Haisnap Emerges, Allowing Even Novice Users to Easily Create AI Applications
Haisnap is a new AI application development tool designed to allow users with zero programming experience to easily create personalized web applications. Users simply describe their needs in natural language, and Haisnap automatically generates the application and provides a complete source code download, enhancing transparency and trust. Its multi-AI agent collaboration mechanism improves development efficiency and is suitable for various application scenarios, from entertainment to utility, promoting the popularization of no-code development.
【AiBase Summary:】
🎨 Haisnap is a low-code or even no-code AI application development platform where users can generate web applications by simply describing their needs in natural language.
💻 Haisnap supports complete source code download, enhancing user trust and providing secondary development opportunities for users with programming experience.
🚀 Haisnap's multi-AI agent collaboration mechanism improves development efficiency and is suitable for various application scenarios, promoting the popularization of no-code development.
Details: https://www.haisnap.com/
10. Alipay Launches "Payment MCP Server" for AI Developers, Enabling Rapid Implementation of In-App Payments for AI Agents
The "Payment MCP Server" service, jointly launched by Alipay and the ModelScope community, provides AI developers with a convenient payment integration method, aiming to bridge the gap between AI services and commercialization. Through the MCP protocol, AI agents can quickly implement payment functions, supporting various payment scenarios on mobile and web ends.
【AiBase Summary:】
💡 Payment MCP Server allows AI agents to integrate payment capabilities with one click, shortening developer integration time.
📱 Supports mobile and web-end payments, meeting diverse intelligent payment needs.
🔄 Provides full-process payment management, including payment, inquiry, and refund functions.
11. Global Artificial Intelligence Market Projected to Reach $368 Billion by 2034
The global artificial intelligence market is rapidly expanding, projected to reach $368.47 billion by 2034, with a Compound Annual Growth Rate (CAGR) of 19.20%. This growth is mainly driven by increased data availability, improved computing power, and increased investment in AI research. The North American market held a 36.92% share in 2024, while the Asia-Pacific region is the fastest-growing market.
【AiBase Summary:】
🌍 The global AI market is projected to reach $368.47 billion in 2034, with a CAGR of 19.20%.
🚀 The North American market held a 36.92% share in 2024, with the Asia-Pacific region being the fastest-growing market.
💻 Deep learning held the largest share of the AI market in 2024, with strong demand in the service sector.
12. XPeng Motors' Self-Developed Turing AI Chip to Enter Mass Production in 2025 for L4 Autonomous Driving
XPeng Motors chairman He Xiaopeng revealed in an interview that the company's self-developed Turing AI chip is expected to enter mass production in the second quarter of 2025, primarily for L4 autonomous driving, AI robots, and flying cars. This chip boasts 40 cores and can run AI large models with up to 30 billion parameters, exceeding the computing power of Nvidia's Orin X, marking a significant breakthrough for XPeng Motors in autonomous driving technology.
【AiBase Summary:】
🔍 XPeng Motors' self-developed Turing AI chip is expected to enter mass production in the second quarter of 2025, primarily for autonomous driving and AI robots.
💪 The Turing chip has 40 cores and can run AI large models with up to 30 billion parameters, outperforming Nvidia's Orin X.
🚗 All future XPeng models will be equipped with the self-developed Turing chip, completely eliminating dependence on Nvidia and further promoting the development of autonomous driving technology.
13. Alibaba's Quark AI Reaches 150 Million Monthly Active Users, Surpassing ByteDance's Doubao
Alibaba's AI application, Quark, reached nearly 150 million monthly active users in March, surpassing ByteDance's Doubao to become the most popular AI application in China. After restructuring, Quark offers various functions such as academic research, document writing, and image generation, reflecting Alibaba's strong competitiveness in the AI market.
【AiBase Summary:】
✅ Quark reaches 150 million monthly active users, surpassing Doubao to become China's most popular AI application.
🔄 Quark, after restructuring, provides various functions such as academic research and document writing, enhancing user experience.
🌐 Alibaba Cloud expands services, actively responding to the AI market's price war, launching new language models.
14. OpenAI Releases GPT-4.1 Prompt Engineering Guide
On April 15th, OpenAI released a prompt engineering guide for its latest language model, GPT-4.1, aiming to help developers build and optimize AI applications more efficiently. The guide details GPT-4.1's features, provides prompt strategies and best practices from basic to advanced, and emphasizes the importance of clear instructions, structured prompts, and avoiding ambiguity.
【AiBase Summary:】
🛠️ GPT-4.1 has significantly improved programming and instruction-following capabilities; developers need to adjust their prompting methods to adapt to its characteristics.
💡 Core prompting principles include clear instructions, providing structure, and avoiding ambiguity, helping developers create effective prompts.
🔍 OpenAI's advanced prompting strategies, such as Few-shot examples and Chain-of-Thought prompting, can guide the model to perform more accurate reasoning and output.
Details: https://github.com/openai/openai-cookbook/blob/main/examples/gpt4-1_prompting_guide.ipynb
