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 AI news, focusing on developers and helping you understand technology trends and innovative AI product applications.
Discover the latest AI products Learn More: https://top.aibase.com/
1. MCP SDK Officially Supports Streaming HTTP, Offering Developers a Brand New Experience
MCP recently announced support for streaming HTTP and released version 1.10.0 of its Typescript SDK, marking its departure from the traditional SSE protocol. Streaming HTTP provides developers with a more flexible and efficient development experience, allowing communication without the need for persistent connections and simplifying server management. Furthermore, streaming HTTP retains the SSE option, offering increased flexibility. Although client-side support is still in its early stages, the upcoming support will further drive the development of remote hosting and improve development efficiency.
【AiBase Summary:】
🌐 MCP introduces streaming HTTP, replacing the traditional SSE protocol for improved development flexibility.
⚙️ The new Typescript SDK 1.10.0 supports streaming HTTP, simplifying server management.
🚀 Client-side support is coming soon, boosting remote hosting development and efficiency.
2. Vidu Q1 Officially Launches, Featuring Higher Definition and Smoother Frame Transitions
Shengshu Technology's Vidu Q1, a high-performance generative AI video model, has quickly garnered industry attention due to its exceptional visual quality, smooth transitions, and accurate sound effects. The comprehensive upgrade of its four core functions provides creators with a professional-level filmmaking experience, marking a new milestone in AI video generation technology. The innovative technology architecture and diverse application scenarios make Vidu Q1 suitable for social media, film advertising, game development, and more.
【AiBase Summary:】
🎨 Superior Image Quality: Supports 1080p video output with detail comparable to professional VFX, ideal for anime character generation.
🎬 Cinematic Transitions: Utilizes "First-to-Last Frame" technology for seamless transitions, supporting complex scene generation.
🔊 Precise Sound Effects: 48kHz high-definition AI sound effect generation, allowing users to customize sound effects and background music to enhance video emotion.
3. Musk's xAI Plans to Raise $25 Billion to Develop Colossus 2 Supercomputer
xAI, the artificial intelligence company founded by Elon Musk, recently announced a new project, planning to raise up to $25 billion to develop its next-generation supercomputer, Colossus 2. This move will further enhance xAI's competitiveness in the AI field and potentially pose a challenge to OpenAI. xAI's current annual revenue is approximately $1 billion, with a valuation between $150 billion and $200 billion. Colossus 2 is expected to support up to 1 million NVIDIA GPUs, with R&D costs ranging from $35 billion to $40 billion.
【AiBase Summary:】
💰 xAI plans to raise up to $25 billion for the development of Colossus 2.
📈 xAI's valuation is expected to be between $150 billion and $200 billion, with an annual revenue of approximately $1 billion.
🖥️ Colossus 2 will support up to 1 million NVIDIA GPUs, with R&D costs estimated at $35 billion to $40 billion.
4. Sand AI Open-Sources MAGI-1 Video Generation Model: Infinitely Extensible, High-Fidelity
On April 21, 2025, Sand AI launched the MAGI-1 open-source video generation model, quickly attracting attention for its autoregressive diffusion architecture and exceptional generative capabilities. The model not only supports high-fidelity video generation but also features infinite extensibility and flexible timeline control, making it suitable for film production and complex narrative scenarios. MAGI-1's open-source nature and strong community feedback have made it stand out in the video generation field, and it is expected to play a larger role in real-time generation and virtual reality applications in the future.
【AiBase Summary:】
🚀 MAGI-1 uses an autoregressive diffusion architecture, supporting efficient video generation with a native resolution of 1440x2568px.
🛠️ The model is open-source and offers Docker deployment support, lowering the barrier to entry for developers.
🌟 MAGI-1 features infinite video extension and second-level timeline control, suitable for creating complex dynamic scenes.
Details: https://github.com/SandAI-org/Magi-1
5. Supercharged AI E-commerce Shopping Assistant! Add To Cart AI: Understands Text, Voice, and Images to Help Users Find Products
Add To Cart AI is an innovative e-commerce shopping assistant designed to enhance the consumer shopping experience through AI technology. It can quickly convert shopping lists, chat logs, and images into complete shopping carts, significantly improving shopping efficiency. In addition, ACAI features personalized recommendations, providing accurate product suggestions based on user history and real-time queries. The tool seamlessly integrates with multiple mainstream e-commerce platforms, ensuring that both merchants and consumers enjoy a convenient shopping experience.
【AiBase Summary:】
🛒 Add To Cart AI can quickly convert shopping lists, chats, and images into complete shopping carts.
🤖 It features powerful personalized recommendations, intelligently recommending products based on user preferences.
🛍️ The tool has successfully integrated with multiple e-commerce platforms, including WooCommerce, Magento, and Shopify.
Details: https://top.aibase.com/tool/add-to-cart-ai
6. Anthropic Releases Claude Code Best Practices Guide to Help Developers Seamlessly Integrate AI into Programming Workflows
Anthropic's recently released Claude Code best practices guide provides developers with a flexible tool designed to seamlessly integrate AI technology into daily programming tasks. The guide emphasizes low-level tools centered around the command-line interface, allowing developers to customize their workflow as needed. Through CLAUDE.md files, tool integration, and multi-agent parallel development, Claude Code offers an efficient and secure coding model to help engineers optimize workflows and improve development efficiency.
【AiBase Summary:】
📄 Claude Code is a command-line integrated development assistant that supports custom CLAUDE.md documentation, improving task adaptability and context awareness.
🔧 The tool is compatible with existing development environments, supporting Unix tools and version control, and can extend functionality through REST APIs, simplifying developer collaboration.
🚀 The guide emphasizes the importance of planning and task decomposition, supporting test-driven development and headless operation to improve automation and parallel development efficiency.
Details: https://www.anthropic.com/engineering/claude-code-best-practices
7. Realme Launches its First AI Translation Earbuds, Bud Air7 Pro, Supporting 32 Languages!
Realme is about to launch its first AI translation earbuds, Bud Air7Pro, powered by iFLYTEK's Starfire cognitive large model 4.0Ultra. It boasts powerful translation capabilities, supporting face-to-face translation and simultaneous interpretation between Chinese and 32 other languages. The earbuds' launch coincides with the Realme GT7 phone launch event, showcasing more innovative technologies. The GT7 phone is equipped with a Dimensity 9400+ chip, a 7200mAh battery, and 100W flash charging, along with the new GT Performance Engine 2.0 to enhance gaming smoothness.
【AiBase Summary:】
🗣️ Bud Air7Pro earbuds support real-time translation between Chinese and 32 languages, greatly improving cross-language communication efficiency.
📱 The Realme GT7 phone is equipped with a Dimensity 9400+ chip, a 7200mAh large battery, and supports 100W flash charging.
🔥 The GT7 phone uses the industry's largest 7700mm² single VC vapor chamber for stability during extended use.
8. Top 20 AI Video Generation Companies of 2025 Announced: Kuaishou's Keling AI, Aiming AI, and PixVerse AI Rank Among the Top Three
DBC, CIW, and CIS recently released the "Top 20 AI Video Generation Companies of 2025" list, showcasing outstanding companies in this field. Kuaishou's Keling AI, Aishitech's PixVerse AI, and Douyin's Aiming AI took the top three spots, demonstrating their leading position in AI video generation technology. Furthermore, several well-known companies such as Alibaba, Tencent, and iFLYTEK also ranked highly, further proving the rapid development and market potential of AI video generation technology.
【AiBase Summary:】
🎥 Keling AI, Aiming AI, and PixVerse AI rank among the top three, showcasing their strong technological capabilities.
🏢 Many well-known companies such as Alibaba and Tencent are also on the list, indicating fierce industry competition.
🚀 AI video generation technology is developing rapidly, with huge future market potential.
9. GLM-4-32B and GLM-Z1-32B Launched on OpenRouter, Freely Available
The GLM-4-32B and GLM-Z1-32B, two cutting-edge large language models developed by Tsinghua University's KEG lab, have been launched on the OpenRouter platform and are completely free to users worldwide. This milestone event marks the popularization of high-performance AI models, providing developers and researchers with powerful tools to drive innovation in AI applications. GLM-4-32B boasts 32 billion parameters and excellent performance, while GLM-Z1-32B excels in solving complex tasks. The release of these two models further solidifies THUDM's leading position in global AI research.
【AiBase Summary:】
🌟 GLM-4-32B is a 32-billion parameter general-purpose large language model suitable for various natural language processing tasks, with performance comparable to top models.
🔍 GLM-Z1-32B introduces an innovative "rumination" mechanism, particularly suitable for handling open-ended complex queries, offering significant research advantages.
🌐 The OpenRouter platform provides a unified API interface, allowing users to quickly integrate models, lowering the barrier to entry and promoting collaboration within the global AI community.
Details: https://openrouter.ai/models
10. Columbia University Dropout Develops "AI Cheating Tool" Interview Coder, Successfully Secures $5 Million in Funding
Recently, Cluely, a startup founded by Columbia University dropouts Chungin "Roy" Lee and Neel Shanmugam, announced the completion of a $5.3 million seed funding round, launching an AI tool called Interview Coder, dubbed an "interview cheating tool." This tool can secretly use artificial intelligence in various scenarios to help users access real-time information. Although the two founders were suspended by the university for developing the tool, their efforts have resulted in Cluely's annual recurring revenue exceeding $3 million, sparking discussions about the ethical use of AI tools in job interviews.
【AiBase Summary:】
💼 Cluely secures $5.3 million in funding, launching the "interview cheating tool" Interview Coder.
🎓 The two founders were suspended from Columbia University for developing the tool and have since officially dropped out.
💰 Lee used the tool to successfully secure an internship at Amazon, with annual recurring revenue exceeding $3 million.
Details: https://www.interviewcoder.co/
11. Enhanced AI Governance Capabilities! Douyin Bans 2.6 Million Black Market Accounts in Q1
Douyin's Q1 2025 black market governance data report shows significant progress in combating black market activities. By building an AI-centric governance system, Douyin banned 2.6 million accounts involved in online water armies, fraud, and illegal traffic redirection, and reported illegal leads to relevant authorities. The deep application of AI technology has significantly improved the platform's efficiency in risk identification and handling, protecting user safety and maintaining a healthy platform ecosystem.
【AiBase Summary:】
🔒 Douyin banned 2.6 million black market accounts involved in online water armies, fraud, and other violations.
🤖 Using AI technology, Douyin has improved the efficiency of black market governance, reducing the processing time for individual cases to seconds.
📉 The number of violation reports decreased by 73.3%, with nearly 1 million illegal traffic redirection accounts banned, maintaining platform security.
12. The World's First Autoregressive Video Generation Large Model MAGI-1 Makes a Major Debut, Led by the Swin Transformer Team, Ushering in a New Era of Video Creation
