Welcome to the AI Daily section! Here, you'll find your daily guide to exploring the world of artificial intelligence. Each day, we bring you the hottest topics in the AI field, focusing on developers, helping you stay on top of tech trends and understand innovative AI product applications.
Discover Fresh AI Products Click to Learn More: https://top.aibase.com/
1、Cerebras Inference: The World's Fastest AI Inference Architecture, 20 Times Faster!
Cerebras Systems' Cerebras Inference is set to revolutionize AI inference, offering 20 times the speed at a significantly lower cost, setting a new benchmark for AI computing. This groundbreaking solution achieves remarkable speed while maintaining industry-leading accuracy, ushering in a new era for AI inference.
AiBase Highlights:
🌟 20 times faster speed, more competitive pricing, ushering in a new era for AI inference.
💡 Supports various AI models, particularly excels in large language models (LLMs).
🚀 Offers three service tiers, allowing developers and enterprise users to choose flexibly.
2、Claude's Powerful Visualization Feature, Artifacts, Now Available for All Users
Anthropic's AI startup has introduced a new feature for the Claude series of large language models called "Artifacts," which allows users to easily generate interactive visualizations, charts, or mini-games, enhancing creative efficiency. The feature is now available in the free, professional, and team versions, supports mobile use, and promotes global creative exchange.
AiBase Highlights:
🔧 Claude Artifacts feature now available in all user's free, professional, and team versions, supports mobile use.
🌍 Only users on the free and professional plans can publish and remix Artifacts, promoting global creative exchange.
📊 Supports multiple output formats, suitable for developers, product managers, designers, marketing, and sales teams, enhancing work efficiency.
3、The Master of Reasoning in AI!OpenAI Shows "Strawberry" Project to the FBI, May Launch in Fall
OpenAI has demonstrated the Strawberry project to U.S. national security officials, promising to develop and use it within safe and reasonable limits. Strawberry will function as part of a chatbot, capable of solving math and programming problems. The company's business is growing rapidly, with a valuation as high as $86 billion.
AiBase Highlights:
🍓 OpenAI's Strawberry project will focus on solving complex math and programming problems.
🚀 The company's business is growing rapidly, with double the revenue from LLM and ChatGPT subscriptions, valued at $86 billion.
🔍 Using Strawberry to generate high-quality training data is expected to significantly reduce errors in AI models.
4、Zhipu Open-Sources CogVideoX-5B Video Generation ModelEven Older Graphics Cards Can Handle It
The ModelScope community has recently announced that the larger version of the domestic open-source Sora video generation model, CogVideoX-5B, is now officially open-source. The new model has significantly improved the quality and visual effects of video generation, adopting 3D causal variational autoencoders and expert Transformer technology, combined with text and video embeddings, using 3D-RoPE as positional encoding, and utilizing 3D full attention mechanisms for spatio-temporal joint modeling. The model also employs progressive training techniques to generate high-quality videos with significant motion features, coherence, and long duration.
AiBase Highlights:
✨ The new CogVideoX-5B model has significantly improved the quality and visual effects of video generation.
🔍 CogVideoX-5B is based on a large-scale DiT model, designed for text-to-video generation tasks.
⚙️ The model employs 3D causal variational autoencoders and expert Transformer technology, combined with text and video embeddings, using 3D-RoPE as positional encoding, and utilizing 3D full attention mechanisms for spatio-temporal joint modeling.
Details Link: https://modelscope.cn/models/ZhipuAI/CogVideoX-5b
5、Quark PC AI Fully Upgraded, Integrating AI Search, AI Creation, AI Summary, and More
The Quark PC application has been upgraded with enhanced AI search, AI writing, AI PPT, AI document summarization, and other key functions. The AI search feature has a more powerful model capability, achieving faster interaction speeds, leading the industry standard. In terms of writing, it supports nearly 200 types of document creation and can quickly produce high-quality articles. System-level full-scene AI capabilities allow users to invoke AI functions through shortcut keys to perform operations such as search, interpretation, translation, and polishing.
AiBase Highlights:
✨ AI search feature upgraded, interaction speed faster, providing accurate answers.
📝 Supports nearly 200 types of document creation, producing high-quality articles in half a minute.
🔍 System-level full-scene AI capabilities, shortcut keys to invoke AI functions for search, interpretation, translation, and polishing.
6、Test Period Free Trial! Metaverse Launches Physics-Based 3D Motion Generation Model, MotionGen
Shenzhen Metaverse Information Technology's MotionGen model is a significant breakthrough in China's 3D AIGC field. By integrating large models, physical simulation, and reinforcement learning algorithms, users can quickly generate realistic and smooth 3D motions through simple text commands. This model lowers the threshold for 3D content production, increases creative freedom and efficiency, and is of great significance to the animation, gaming, film, and virtual reality industries.
AiBase Highlights:
⚙️ The MotionGen model greatly lowers the threshold for 3D content production, allowing zero-experience creators to easily create high-quality animations.
🤖 MotionGen has precise text understanding capabilities, supports various complex motion creative needs, and achieves seamless text-to-motion connections.
🌟 MotionGenGPT algorithm combines multiple technologies to enhance the detail representation and matching of motion generation.
7、Zhipu AI Opens GLM-4-Flash: Free to Use, Multi-Scenario Adaptation
The Zhipu AI open platform recently announced that the GLM-4-Flash model is now free to use, leading a new wave of innovation in the AI application field. The model is fast and economical, suitable for various tasks, supports multiple languages, and optimizes running efficiency and reduces inference costs. It has a wide range of applications and outstanding performance, showing potential in multiple fields. Users can obtain free fine-tuning benefits by completing a survey questionnaire.
AiBase Highlights:
⚙️ GLM-4-Flash model is now free to use, suitable for simple tasks, supports multiple languages, optimizes running efficiency and reduces inference costs.
💡 GLM-4-Flash has outstanding performance, fast generation speed, and real-time web retrieval and function call capabilities.
🌐 GLM-4-Flash has a wide range of applications, showing potential in areas such as biological data processing, training data generation, AI tutor creation, and translation app development.
Details Link: https://bigmodel.cn/console/trialcenter?modelCode=glm-4-flash
8、Google Meet Introduces AI Auto-Notes Feature, No More Missed Information in Meetings
Google Meet has introduced a new feature called "Take Notes for Me," making meeting notes more convenient and efficient. This feature not only automatically summarizes meeting content but also provides a meeting summary for latecomers, allowing everyone to keep up with the meeting progress. Additionally, the meeting notes are saved as Google Docs and automatically sent to participants, providing a one-stop service. Although the feature is powerful, it currently only supports English and the AI may make mistakes. It is expected to be rolled out to all Google Workspace users by September 10, 2024.
AiBase Highlights:
📌 Automatically summarizes meeting content, improving meeting note efficiency.
🕒 Provides meeting summaries for latecomers, helping to quickly catch up with the meeting progress.
📄 Meeting notes are saved as Google Docs and automatically sent to participants, providing convenient services.
9、AI Reshapes Mythology: Douyin Blogger Creates "Classic of Mountains and Seas" Visual Feast in 8 Days, Netizens Amazed Beyond CCTV
In China's rich cultural treasure trove, "Classic of Mountains and Seas" is undoubtedly a brilliant gem. Douyin blogger AI Madhouse used artificial intelligence technology to complete the AI-recreated "Classic of Mountains and Seas" video in just 8 days, accurately restoring the magical world in the original work, sparking heated discussions and receiving praise from the audience. This new form of creation showcases the powerful potential of AI in video production, bringing more possibilities to content creation.
AiBase Highlights:
🌟 Douyin blogger AI Madhouse reshapes "Classic of Mountains and Seas" with AI technology, accurately restoring the magical world, the video quality is amazing.
🔮 The video was completed in just 8 days, showing the potential of AI in short video production, greatly reducing production costs and cycles.
🎮 The audience is full of praise, some netizens believe the video quality surpasses CCTV, showing the potential of "Classic of Mountains and Seas" IP adaptation into games.
10、AI Deepfake Chaos: Trump Falls for AI Fake Image, Musk Involved in Fraud Cases
Recently, Trump shared a fake image, causing Taylor Swift fans to boycott, and Musk was involved in an AI deepfake fraud incident. The incident warns that investment and information acquisition need to be cautious, and in the digital age, one must guard against becoming a victim of false information.
AiBase Highlights:
🤖 Trump shared a fake image, causing Taylor Swift fans to boycott.
🔍 Musk was involved in an AI deepfake fraud incident, victims lost hundreds of thousands of dollars.
😱 In AI deepfake fraud cases, Musk is one of the most common spokespersons.
11、Google Releases New Version of Gemini 1.5 Series Models
Google has recently released the latest version of the Gemini 1.5 series models, including Gemini 1.5Flash-8B, Gemini 1.5Flash, and Gemini 1.5Pro, with significant performance improvements. Developers are excited about the comprehensive performance enhancements of Gemini 1.5Flash. However, user feedback on the new model is polarized, with some looking forward to the comprehensive upgrade of Gemini 2.0, while others appreciate the speed and performance of the current updates.
AiBase Highlights:
✨ Gemini 1.5 series models have significant improvements in mathematics, programming, and complex prompts.
🚀 Gemini 1.5Flash can handle multi-modal inputs of over 10 million tokens, with strong adaptability.
🛠️ User feedback is polarized, some looking forward to the comprehensive upgrade of Gemini 2.0, others appreciating the current updates.
12、Small but Beautiful! HKU's Latest Recommendation System, EasyRec, Gains Insights into User Preferences through Text
In an era of information overload, recommendation systems have become important assistants in filtering information for us. The emergence of EasyRec may solve the problem of mismatched content and insufficient accuracy, as its language model-based recommendation system performs excellently in predicting user preferences.
