Welcome to the 【AI Daily】 section! This is your guide to exploring the world of artificial intelligence every day. Here, we present you with hot content in the AI field daily, focusing on developers to help you gain insights into technological trends and understand innovative AI product applications.
Fresh AI products click to learn more: https://top.aibase.com/
1. 🌟 OpenAI Announces Free Memory Function for All ChatGPT Users
OpenAI has updated its support documentation, announcing that the memory function will be made free for all ChatGPT users, including logged-in free users, enhancing personalized conversational experiences.
【AiBase Summary:】
💬 The memory function supports short-term conversation continuity; free users can experience the basic version.
💰 Paid users can reference longer-term conversation records, enjoying higher convenience and depth of interaction.
🔒 Users can manage their memory settings, close or delete specific memory content at any time, protecting privacy.
2. Ant Group Launches "AI Health Butler," Serving Over 40 Million Users with Intelligent Health Services
Ant Group's "AI Health Butler" has passed the Trusted Evaluation of the Telecommunications Technology Institute (信通院) in the healthcare industry, becoming one of the first products to pass the evaluation. This marks Ant Group's continuous exploration in the medical AI field and showcases its leading position in the healthcare sector.
【AiBase Summary:】
🌟 The AI Health Butler has passed the trusted evaluation by the Telecommunications Technology Institute, ensuring safety and effectiveness.
👥 It has served over 40 million users, with over 60 well-known doctors' AI entities joining in.
🚀 Provides personalized services such as doctor appointment booking, health assessment, and medical report interpretation.
3. Anthropic Offers a Free Course on Building AI Applications Using MCP
Anthropic has collaborated with DeepLearning.AI to launch the free course 'MCP: Building Rich Context AI Applications with Anthropic', helping developers master the MCP protocol and simplify the connection between AI applications and external tools and data.
【AiBase Summary:】
🌟 MCP is a universal protocol that enhances context processing capabilities by standardizing the interaction between LLMs and external data sources.
📚 The course covers core concepts, architecture, and practical projects of MCP, helping developers quickly get started and build intelligent AI applications.
🌐 MCP is open source and supports integration with multiple tools and data sources, promoting AI development standardization and cross-domain integration.
More details: https://www.deeplearning.ai/short-courses/mcp-build-rich-context-ai-apps-with-anthropic/
4. Google DeepMind Introduces New Technology: Generating Realistic Motion Videos Without 3D Models
The DeepMind team collaborated with Brown University to develop the 'Force Prompting' technology, which can generate realistic motion effects without using 3D models or physical engines. Users can control AI-generated video content by specifying the direction and intensity of forces.
【AiBase Summary:】
🌟 Force Prompting technology can generate realistic motion videos without 3D models or physical engines, just through text instructions.
⚙️ Users can operate by force direction and intensity, achieving natural and smooth motion performance and enhancing video realism.
📈 The model has strong generalization ability, adapting to new scenarios and objects, and even mastering some physical rules.
For more details: https://force-prompting.github.io/
5. Over 400 AI Models Unlock Network Search! Exa Joins Forces with OpenRouter to Ignite RAG Revolution
Exa has partnered with OpenRouter to provide real-time network search functionality for over 400 large language models, enhancing the models' information acquisition capabilities through RAG technology and revolutionizing AI interaction experiences.
【AiBase Summary:】
✨ Exa and OpenRouter collaborate, integrating over 400 large language models with real-time network search, significantly improving information acquisition capabilities.
🔍 Based on RAG technology, models can dynamically access the latest network information, breaking through traditional knowledge update limitations.
💻 Developers can flexibly call via OpenRouter, reducing development costs and expanding AI application scenarios.
6. China National Knowledge Infrastructure (CNKI) Launches CNKI AI
China National Knowledge Infrastructure (CNKI) has launched CNKI AI, a new exploration result based on artificial intelligence technology. The platform combines an AI academic research assistant with enhanced retrieval, providing more precise and comprehensive knowledge services.
【AiBase Summary:】
✨ Provides question-and-answer enhanced retrieval and generative knowledge services, assisting academic research and scientific innovation.
📚 High-quality data and credible controllable characteristics ensure efficient, precise, and reliable service.
🔍 Dual-path retrieval and paragraph retrieval enhance recall and precision rates, meeting diverse needs.
For more details: https://www.wjx.cn/vm/eikFgVh.aspx
7. Anthropic Launches Claude Explains Blog Project to Explore New Collaborative Models Between AI and Human Experts
Anthropic has launched the 'Claude Explains' blog project to showcase the capabilities of its AI model Claude in content creation. The blog content is generated by Claude AI and edited by human experts; the initial articles focus on technical themes, with plans to expand into more fields in the future.
【AiBase Summary:】
📌 Claude AI generates content, while human experts edit it to enhance professionalism and readability.
🌟 The blog covers technical topics, simplifying complex code libraries to assist technical development.
🌐 Anthropic plans to expand the scope of topics to include creative writing, data analysis, and more fields.
8. Claude Pro Upgrades Key Features: Research Mode and Remote MCP Integration Fully Available!
Anthropic announced that Claude Pro has added research mode and remote MCP integration features, enhancing the practicality and productivity of the AI assistant.
【AiBase Summary:】
Research mode compresses complex research tasks from hours to minutes, significantly increasing efficiency.
Remote MCP integration allows Claude Pro users to seamlessly connect various tools, simplifying cross-platform collaboration.
The upgraded Claude Pro is more competitive in terms of functions and pricing, attracting more users.
9. Fish Audio Launches OpenAudio S1: A Super-Natural Voice Model Driven by 2 Million Hours of Data
Fish Audio's OpenAudio S1 is highly anticipated, this text-to-speech model trained on massive amounts of data not only excels in voice naturalness and emotional expression but also offers flexible dual-version options, providing efficient and economical voice generation solutions for businesses and developers.
【AiBase Summary:】
🎤 The model is trained on 2 million hours of audio, supporting diverse language styles and emotional expressions.
🚀 Provides two versions: S1 (4 billion parameters) and S1-mini (500 million parameters), meeting different scene requirements.
🌟 Uses RLHF technology to generate emotionalized voices, enhancing user experience and reducing costs.
10. OpenAI Codex Upgrade: Speech Input and Networking Functions Make Programming Smarter
OpenAI has comprehensively upgraded its programming tool Codex, adding speech input and networking capabilities while lowering usage thresholds and improving developer coding efficiency.
【AiBase Summary:】
The new networking function enables Codex to automatically complete environment configuration, code checking, and testing, allowing developers to focus on logic and functional implementation.
The speech input function allows developers to convey commands more naturally, improving tool usability.
Codex is now available to ChatGPT Plus users, lowering the usage threshold and benefiting more developers.
11. OpenAI Upgrades AI Agent Development Tools, Supports TypeScript and Improves Voice Interaction
OpenAI has made significant upgrades to its AI agent development tools, including adding TypeScript support, optimizing the voice interface, enhancing observability, and improving voice-to-voice models.
【AiBase Summary:】
🌟 TypeScript Support: Agents SDK adds support for TypeScript, enabling JavaScript and Node.js developers to participate in intelligent agent development.
🎤 RealtimeAgent Functionality: Supports low-latency voice applications, allowing suspension of execution and manual confirmation of agent status, suitable for regulatory scenarios.
🔍 Improved Voice Models: Optimize voice-to-voice models, reduce latency, improve conversational naturalness and interruption handling capabilities.
12. Huawei WATCH 5 Smartwatch Integrates Dual Large Models, Upgrading Sports and Health Experience
Huawei officially released the WATCH 5 smartwatch, integrating PanGu large models and DeepSeek large models, achieving breakthrough improvements in voice interaction, health monitoring, and ecological interconnection.
【AiBase Summary:】
⌚️ WATCH5 supports dual AI large models, making voice interaction more convenient and health data analysis more accurate.
🏃♀️ The wrist Mini-Art function analyzes over 200 indicators in 20+ sports and health fields, providing personalized guidance.
🔗 Supports ecological interconnection, compatible with Huawei devices and other third-party health management platforms, promoting comprehensive healthy living.
13. DeepSeek May Have Used Google Gemini Data to Train New AI Models
DeepSeek's recently updated R1 inference AI model has performed excellently in various benchmark tests, but its training data source has sparked controversy. Some developers have pointed out that its model bears similarities to Google Gemini series, and DeepSeek has previously been accused of using "data distillation" techniques to train models. Despite this, the AI community generally believes that this phenomenon may stem from mutual imitation among models.
【AiBase Summary:】
.DeepSeek R1 model performs excellently in math and programming tests, but its training data source is questionable.
Multiple developers have noted that DeepSeek models share similar wording and expression styles with Google Gemini series.
OpenAI found that DeepSeek may have obtained training data through "data distillation" technology, violating relevant rules.
14. Panasonic Launches "OmniFlow" Multimodal Generation AI for Free Conversion Between Text, Image, and Audio
Panasonic Holdings Corporation, in collaboration with UCLA researchers, has developed "OmniFlow," a multimodal generation AI with "any-to-any" generation capabilities, enabling free conversion between text, image, and audio, reducing data collection costs and enhancing generation efficiency.
【AiBase Summary:】
✨ Innovative technology 'OmniFlow' supports free conversion between text, image, and audio, greatly enhancing the application potential of multimodal generation AI.
📚 Requires 1/60th of the data demand compared to traditional methods, significantly reducing data collection costs and optimizing model training efficiency.
🌟 Performs best in text-to-image and text-to-audio tasks, showcasing optimal performance, and is expected to be applied in factories and lifestyle areas in the future.