AI Daily: Xiaohongshu First Releases AI Governance Principles; Honor YOYO First Integrates DeepSeek-V4; Lingguang App First Brings World Models to Mobile Devices

Welcome to the "AI Daily" section! This is your guide to exploring the world of artificial intelligence every day. Every day, we present you with the latest content in the AI field, focusing on developers to help you understand technical trends and innovative AI product applications.

Fresh AI products Click to learn more:https://app.aibase.com/zh

1. Xiaohongshu First Releases its "AI Governance Principles": Resisting AI Infringement, AI Fraud, and AI Impersonation

Xiaohongshu first released its "AI Governance Principles," emphasizing the positive role of AI in creation, while clearly opposing negative AI behaviors such as AI fraud and infringement, committed to maintaining the authenticity and safety of the community.

AiBase Summary:

🤖 Xiaohongshu emphasizes the important role of AI in creation, while clearly opposing negative AI behaviors.

📌 The platform calls on users to truthfully identify AI-related content to ensure the authenticity and transparency of information.

🚫 Xiaohongshu firmly resists AI fraud and infringement, and continues to build capabilities for identifying and governing AI content.

2. Android Launch: Honor YOYO is the First to Integrate DeepSeek-V4 Large Model

The Honor YOYO smart assistant has officially integrated the DeepSeek-V4 large model, marking further breakthroughs in the AI field for the Android ecosystem. Through improvements in performance, context understanding, and reasoning efficiency, YOYO can handle complex tasks more accurately, providing users with a smoother interaction experience.

AiBase Summary:

🧠 Deeply integrated with the DeepSeek-V4 large model, improving the logic and response speed of the smart assistant.

🔄 Optimizing the underlying algorithm, enhancing the ability to process complex instructions and long-text conversations.

📱 Users need to meet specific version requirements to experience the interactive changes brought by the new generation of large models.

3. NetEase Youdao Notes Launches "LLM Wiki" Suite, Redefining Knowledge Management in the AI Era

The "LLM Wiki" skill suite launched by NetEase Youdao Notes marks a new stage in knowledge management in the AI era. By using large language models to actively compile knowledge, it improves the efficiency of information processing and provides individuals and enterprises with an intelligent second brain technology foundation.

AiBase Summary:

🧠 Using large language models (LLMs) to deeply associate and process fragmented knowledge, achieving structured organization of knowledge.

🔄 Shifting from "passive retrieval" to "active compilation," improving the efficiency and value of information processing.

💡 AI capabilities are being embedded, and software forms will focus more on backend logic reshaping, creating an intelligent second brain.

4. Upload One Picture to Instantly Generate a 3D World, Lingguang App Pioneers Bringing World Models to Mobile Devices

Lingguang App enables users to explore a 3D world by simply uploading one picture through its "Experience World Model" feature, while using a mobile game joystick control method to achieve a zero-barrier immersive experience. This feature is the first AGI product in the industry that allows users to experience world models on mobile devices, marking further breakthroughs in AI technology on mobile platforms.

AiBase Summary:

📱 Upload one picture to generate a 3D world and explore it on mobile devices.

🕹️ Introducing a mobile game joystick control method to provide a zero-barrier immersive experience.

🚀 Lingguang App is the first to deploy world models on mobile devices.

5. Xiaomi Opensource VLA Large Model Post-Training Process, Robot Sub-Millimeter-Level Operation Becomes Reality

Xiaomi has made significant breakthroughs in robot technology, solving the challenges of moving robots from laboratories to practical production through the post-training process of the open-sourced VLA large model Xiaomi-Robotics-0. The model demonstrates excellent sub-millimeter-level operation capability, providing strong support for the application of intelligent robots in complex scenarios.

AiBase Summary:

🧠 The Xiaomi-Robotics-0 model enhances the performance of robots in complex tasks through the post-training process.

🔧 The model successfully achieves sub-millimeter-level fine operation capability, demonstrating strong spatial perception and real-time feedback capability.

🌐 Xiaomi has open-sourced technical reports, model weights, and source code, promoting the development of the developer ecosystem and reducing the threshold for high-performance robot development.

Details Link: https://robotics.xiaomi.com/xiaomi-robotics-0.html#pack-earbuds

6. OpenClaw Releases New Version: DeepSeek V4Flash Officially Becomes System Default Model

OpenClaw released a new version, with DeepSeek V4Flash officially becoming the system default model, strengthening multi-modal collaboration and long-chain task stability, and completing the transition from dialogue products to system-level workflow platforms.

AiBase Summary:

🧠 DeepSeek V4Flash becomes the default model of OpenClaw system, enhancing reasoning speed and logic capability.

🔄 Strengthening multi-modal collaboration and long-chain task stability, achieving full-process automation.

🔧 Reconstructing model loading logic and SDK interfaces, completing the transformation from dialogue products to system-level workflow platforms.

7. Hot! Meituan's "Xiaotuan" AI Suddenly Appears on Homepage: The "Lazy Person's Tool" for May Day Travel Has Arrived

The article introduces Meituan's upgrade of its artificial intelligence assistant "Xiaotuan," moving its access point to the App homepage to better serve consumers' needs during the May Day holiday. Through natural language processing and dynamic data verification, "Xiaotuan" can provide personalized recommendations and accurate information, enhancing users' holiday consumption experience.

AiBase Summary:

✨ Upgraded "Xiaotuan" is fully launched, allowing users to directly access it from the homepage.

💡 "Xiaotuan" can understand ambiguous intentions and provide personalized recommendation plans.

🔒 Real-time dynamic verification of merchant information ensures the reliability and accuracy of recommendations.

8. OpenAI Adjusts Strategic Focus: Programming Model Codex Officially Integrated into GPT-5.5 Architecture

OpenAI adjusted its strategic focus, officially integrating the Codex programming model into the GPT-5.5 architecture, reflecting its shift from specialized plug-in to internal comprehensive models, while enhancing the model's efficiency and performance.

AiBase Summary:

✅ OpenAI announced the integration of the Codex programming model into the GPT-5.5 main model, marking a shift in its R&D approach from "specialized plug-in" to "internal comprehensive."

💡 The integrated GPT-5.5 shows significant optimization in resource utilization when handling complex programming tasks, improving developers' cost and efficiency.

🚀 The integration of Codex indicates that AI programming no longer requires independent maintenance but has become a foundational indicator of the intelligence level of general large models.