Welcome to the "AI Daily" column! 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 technology trends and learn about innovative AI product applications.
Fresh AI products click to learn more:https://top.aibase.com/
1. Tencent AudioGenie Makes a Stunning Debut! One-click Generation of Movie-level Sound Effects, Claude and Gemini Tremble!
Tencent AudioGenie, with its powerful multimodal audio generation capabilities and an innovative training-free framework, is redefining the standards for AI audio generation. Facing competition from international giants, AudioGenie demonstrates the solid strength of Chinese AI technology.
AiBase Highlights:
🎥 Supports multiple modal inputs such as video, text, and images, generating audio outputs such as sound effects, voice, and music.
⚙️ Uses a training-free multi-agent framework, achieving efficient collaboration and self-correction through a two-tier architecture.
📈 Performs well in the MA-Bench benchmark test, challenging the market position of Claude and Gemini.
Details link: https://audiogenie.github.io/
2. Alibaba Launches the Multimodal Deep Research Agent WebWatcher
The Alibaba Natural Language Processing team has launched an open-source multimodal deep research agent called WebWatcher, aiming to break through the limitations of existing closed-source systems and open-source agents in the field of multimodal deep research. By integrating various tools such as web browsing, image search, code interpreter, and internal OCR, WebWatcher can handle complex multimodal tasks like a human researcher.
AiBase Highlights:
🌍 WebWatcher is an open-source multimodal deep research agent that can handle complex multimodal tasks.
🧠 Integrates various tools, such as web browsing and image search, to achieve strong visual understanding and logical reasoning abilities.
🚀 In multiple evaluations, WebWatcher's performance significantly outperforms other mainstream models, demonstrating its outstanding capabilities.
Details link: https://github.com/Alibaba-NLP/WebAgent
3. Hong Kong University, Harbin Institute of Technology, and Zhejiang University jointly launch OmniPart, a decoupled 3D model technology, reshaping creative design
The OmniPart technology jointly launched by Hong Kong University, Harbin Institute of Technology, and Zhejiang University brings an important breakthrough to the field of 3D modeling, enabling the independence and structural clarity of 3D model components, significantly improving the accuracy and flexibility of 3D modeling, and applicable to multiple creative fields such as game development and animation production.
AiBase Highlights:
🧠 The OmniPart technology achieves the independence of 3D model components, making creative design more flexible.
🔍 It uses a two-stage generation framework combining autoregressive models and component masks to improve the accuracy of 3D modeling.
🚀 Innovative mechanisms such as voxel dropout enhance the model's effectiveness in complex scenarios.
Details link: https://omnipart.github.io/
4. Meta releases DINOv3, a general image processing AI model without labeled data
META's DINOv3 is a general image processing AI model that does not require labeled data. It is trained using 1.7 billion images through self-supervised learning and has 7 billion parameters. The model performs well in multiple image tasks and fields, especially suitable for professional areas such as satellite image processing.
AiBase Highlights:
🧠 DINOv3 is trained using 1.7 billion images through self-supervised learning, requiring no labeled data.
🚀 With 7 billion parameters, it can handle various image tasks and fields, performing better than the previous model DINOv2.
🌐 Meta has released multiple pre-trained model variants and code on GitHub, allowing commercial use.
Details link: https://github.com/facebookresearch/dinov3
5. China's first legal vertical large model "Xiao Bagong" is launched: traceable and verifiable
China's first legal vertical large model "Xiao Bagong" was officially launched, marking the transition of legal artificial intelligence from academic exploration to large-scale application. The model provides traceable and verifiable legal basis by integrating a large amount of legal data and advanced technology, helping to alleviate the uneven distribution of legal service resources and showing demonstration effects in multiple key areas.
AiBase Highlights:
⚖️ China's first legal vertical large model "Xiao Bagong" was launched, marking the entry of legal artificial intelligence into the stage of large-scale application.
🔍 "Xiao Bagong" integrates 200 million court rulings and 4.2 million laws and regulations, and has the ability to accurately exclude "layman's concepts".
💡 Legal AI has the potential to alleviate the uneven distribution of legal service resources and promote digitalization and inclusiveness in fields such as administrative reconsideration and procuratorial supervision.
6. ChatGPT mobile app revenue breaks 2 billion dollars, crushing competitors creating a 30 times revenue gap
The ChatGPT mobile application achieved remarkable revenue performance in the global market, far exceeding other competitors. Its revenue growth is rapid, with significant increases in user downloads and consumption levels, showing its dominant position in the AI assistant field.
AiBase Highlights:
(ChatGPT mobile application revenue reached 2 billion dollars, 30 times that of competitors)
(ChatGPT monthly revenue growth reached 673%, far surpassing other chatbots)
(ChatGPT global downloads reached 690 million times, 17 times that of Grok)
7. Android phones collectively copy the "Dynamic Island", new chip power doubles, promoting the full explosion of AI functions
The article points out that Android manufacturers have all drawn inspiration from Apple's "Dynamic Island" interaction method in system design and optimized it according to their own characteristics. At the same time, the enhanced computing power of the new generation of chips provides hardware support for the popularization of AI functions, with AI functions integrated into systems by manufacturers to provide a more intelligent service experience.
AiBase Highlights:
✨ Android manufacturers have introduced interaction designs similar to Apple's "Dynamic Island", enhancing user experience.
⚡ The computing power of the new generation of chips has doubled, laying the foundation for the popularization of AI functions.
🤖 Manufacturers fully integrate AI functions, providing intelligent services such as one-click ticket booking and itinerary planning.
8. European AI startup launches chicken brain and fly brain models, 94MB ultra-small AI can run offline on Apple Watch
European AI startup Multiverse Computing released two extremely small AI models, named SuperFly and ChickBrain. These models are compact and can run locally on IoT devices, smartphones, tablets, and personal computers without internet connection. They perform exceptionally well, even surpassing the original models in some benchmark tests.
AiBase Highlights:
✨ Multiverse Computing has launched two ultra-small AI models suitable for various devices and supporting local operation.
🧠 SuperFly and ChickBrain models are named after fly brains and chicken brains, with powerful features and reasoning capabilities.
💰 The company further promotes its quantum-inspired compression technology through 189 million euros in funding and collaborates with several major companies.
9. Major update to Claude Code! New programming tutor mode, beginners can also enjoy one-on-one code guidance
Anthropic launched an important feature update to Claude Code, adding a personalized communication style setting for programming beginners. Users can customize the communication method through commands, including explanatory and learning styles, to meet different learning needs.
AiBase Highlights:
🧠 Explanatory style focuses on in-depth teaching, helping developers understand the principles behind the code.
👩🏫 Learning style adopts an interactive teaching approach, enhancing users' hands-on ability and independent problem-solving skills.
🌐 The newly added programming tutor mode allows beginners to enjoy one-on-one code guidance, lowering the learning barrier.
10. AI technology is abused as a "refund tool", merchants helpless: fake pictures are too realistic, have no way to complain
The article points out that e-commerce platforms have seen the use of AI to forge images of damaged goods for malicious refunds, seriously harming merchants' interests. Legal experts believe this behavior may be illegal and call for strengthened regulation and technological innovation to address this issue.
AiBase Highlights:
🤖 AI tools are used to forge images of damaged goods to claim refunds.
⚖️ Malicious refund behavior may constitute civil fraud or criminal fraud.
🔒 Merchants need to optimize after-sales processes and retain evidence to protect their rights.
11. IDC report: 2024 China AI public cloud service market surges, Alibaba Cloud retains the top position in the Chinese market
IDC reports that the China AI public cloud service market grew rapidly in 2024, mainly driven by the expansion of generative AI applications and the rising demand for machine learning. Sub-sectors such as computer vision, conversational AI, and natural language processing showed strong performance. At the same time, technology providers need to focus on AI governance and cloud architecture optimization to adapt to the demands of the intelligent era.
AiBase Highlights:
🧠 The China AI public cloud service market is expected to reach 19.59 billion yuan in 2024, growing by 55.3% year-on-year.
🖼️ The computer vision and conversational AI markets performed well, reaching 8.1 billion yuan and 2.09 billion yuan respectively.
🛠️ Technology providers need to restructure cloud service architectures, strengthen AI governance to ensure transparency and compliance.
