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.
Explore New AI Products Learn More: https://top.aibase.com/
1. Kunlun Wanwei Open-Sources Skywork R1V Visual Reasoning Chain Model
Kunlun Wanwei has launched Skywork R1V, the world's first open-sourced multi-modal reasoning model. Boasting 3.8 billion parameters, its performance is close to the well-known closed-source model DeepSeek-R1. R1V excels in visual question answering and complex reasoning tasks, achieving impressive scores of 69 and 67.5 on the MMMU and MathVista benchmarks respectively.
【AiBase Summary:】
🌟 The world's first industry-open-sourced multi-modal reasoning model, Skywork R1V, is officially released with 3.8 billion parameters.
🚀 R1V demonstrates outstanding performance in multiple benchmark tests, particularly achieving high scores of 69 and 67.5 in MMMU and MathVista respectively.
📚 Kunlun Wanwei's open-source initiative aims to promote technology sharing, injecting vitality into the global AI open-source community and contributing to the realization of AGI.
Details: https://huggingface.co/Skywork/Skywork-R1V-38B
2. Doubao AI Programming Capabilities Upgraded: HTML Preview and More
Doubao recently significantly upgraded its AI programming features on its web and desktop versions to enhance user programming efficiency and experience. These upgrades include real-time HTML preview, direct Python code execution, and the ability to generate complete project code. Users can more intuitively develop web pages and small games, quickly fix Python code errors, and easily generate complete project code, simplifying the development process. These new features significantly improve user convenience and efficiency during programming.
【AiBase Summary:】
🌐 Added real-time HTML preview, allowing users to intuitively create small games and web pages, enhancing the development experience.
🐍 Supports direct execution of Python code, with AI automatically fixing errors, reducing debugging time.
📦 Added the ability to generate complete project code, simplifying the generation of front-end and back-end logic, improving development convenience.
3. Google Gemini Introduces "Canvas" and Audio Overview Features for Enhanced Collaboration
Google recently launched Gemini's new "Canvas" feature to improve user creation and collaboration. This feature allows users to easily edit and share writing and programming projects, providing a more efficient collaborative approach. With Canvas, users can update drafts in real-time and generate code previews. Additionally, an audio overview feature has been introduced, allowing users to generate audio summaries of documents. These new tools make Gemini a more powerful creative partner, greatly facilitating user workflows.
【AiBase Summary:】
📝 The Canvas feature allows users to easily draft and edit long-form information in Gemini, supporting real-time updates and collaboration.
💻 Provides programming tools; users can generate and preview HTML, React code, and view the results in real-time.
🎧 Added audio overview functionality, allowing users to quickly generate audio summaries of documents for easy sharing and downloading.
4. Cursor Launches Claude Max, Reshaping AI Programming
Cursor's newly launched Claude Max model redefines the standard for AI-assisted programming with its superior performance and innovative capabilities. The model boasts exceptional context processing capabilities, handling up to 200,000 words at once, enabling developers to manage entire project codebases more efficiently. Simultaneously, Claude Max possesses powerful tool invocation and code comprehension capabilities, significantly improving programming efficiency.
【AiBase Summary:】
🚀 Claude Max has a context processing capability of up to 200,000 words, allowing developers to input entire project codebases at once.
⚙️ Supports up to 200 tool invocations, significantly improving the efficiency of editing and optimizing code.
💰 Charged based on usage, suitable for advanced users handling complex projects, rather than everyday coding tasks.
5. Adobe Unveils 10 AI Agents, Enabling Personalized Website Creation
Adobe is once again leading the generative AI wave, launching 10 new AI agents designed to enhance customer experience. These agents span customer interaction, content creation, data management, and more, working together to help businesses more effectively manage customer relationships and optimize websites. Simultaneously, Adobe introduced Brand Concierge, a new feature providing personalized website access experiences, further boosting customer engagement and loyalty.
【AiBase Summary:】
🤖 Adobe launches 10 AI agents to improve customer interaction and content production efficiency.
🌐 The new Brand Concierge feature provides a personalized website experience, enhancing customer engagement.
📈 Generative AI traffic is significantly increasing on retail and travel websites, showing increased consumer acceptance of AI experiences.
6. ByteDance's Doubao Large Model Team Holds All-Hands Meeting, Exploring New AI Heights
Amidst the rapid development of artificial intelligence, ByteDance's Doubao large model team held an all-hands meeting to define its future direction. Co-hosted by Zhu Wenjia and Wu Yonghui, the meeting emphasized the importance of exploring the upper limits of intelligence and encouraged team members to participate in challenging research. Wu Yonghui also proposed increasing resource investment in the Seed Edge project to attract and cultivate top talent.
【AiBase Summary:】
🚀 The Seed team's primary goal is to explore the boundaries of intelligence, conducting in-depth research around the AGI research plan.
💡 Zhu Wenjia encourages the team to participate in AI research with uncertainty, emphasizing the importance of challenging topics.
🌍 The team plans to open-source smaller Dense models to promote technology application and external collaboration.
7. Stability AI Releases New Model: Stable Virtual Camera, Easily Converting 2D Photos to 3D Videos
Stability AI's Stable Virtual Camera is an innovative AI model that transforms 2D images into immersive videos, providing realistic depth and perspective. The model allows users to generate new viewpoints from one or more images, specifying camera angles and supporting various dynamic effects. However, the current version is still a research preview and may experience quality degradation in specific scenarios.
【AiBase Summary:】
🌟 Stable Virtual Camera converts 2D images into immersive videos, offering multiple camera path options.
📉 The current model is a research preview, and quality degradation may occur when processing certain scenes.
💼 After experiencing management crises, Stability AI is actively restructuring and launching new products to improve its prospects.
Details: https://top.aibase.com/tool/stable-virtual-camera
8. 100 Quintillion Calculations Per Second! Nvidia Unveils Two Personal AI Supercomputers: DGX Spark and DGX Station
At the 2025 Global Technology Conference, Nvidia's founder and CEO Jensen Huang unveiled two groundbreaking personal AI supercomputers, DGX Spark and DGX Station. These devices not only boast incredible computing power, achieving up to 100 quintillion AI calculations per second, but also offer new possibilities for innovation in the edge computing field.
【AiBase Summary:】
⚡ DGX Spark boasts 100 quintillion AI calculations per second, utilizing the GB10 Grace Blackwell superchip, suitable for complex AI model processing.
🖥️ DGX Station is equipped with the GB300 Grace Blackwell Ultra Desktop superchip and 784GB of memory, providing an exceptional desktop computing experience.
🌐 Nvidia's two supercomputers aim to support edge computing, helping businesses quickly achieve AI model prototyping and tuning.
9. Nvidia Introduces New Dynamo Software, Aiming to Boost DeepSeek AI Speed by 30x
At the March 18th GTC conference, Nvidia CEO Jensen Huang announced the launch of Dynamo software, aiming to increase the AI processing speed of DeepSeek by 30 times. This move addresses the market disruption caused by DeepSeek's R1 AI program. Dynamo software can distribute AI inference tasks to up to 1000 GPUs for parallel processing, significantly increasing query throughput. Service providers can process customer queries more efficiently, leading to increased revenue.
【AiBase Summary:】
🌟 Nvidia introduces Dynamo software, significantly boosting DeepSeek AI processing speed.
💰 Service providers can process customer queries more efficiently through Dynamo, increasing overall revenue.
🖥️ The new Blackwell "Ultra" chip and DGX Spark computer were officially launched at the conference.
10. Grok Launches DeeperSearch, Enhancing Real-time AI News Retrieval
Recently, xAI's AI assistant Grok added the DeeperSearch feature, significantly improving its ability to retrieve real-time AI news on Twitter. This feature can quickly analyze trending topics from the past 48 hours, with positive user feedback demonstrating Grok's powerful information processing capabilities.
【AiBase Summary:】
📰 DeeperSearch analyzes trending AI news on Twitter over the past 48 hours, providing timely information.
🚀 Grok's upgrade enhances its real-time data processing capabilities, particularly excelling on high-velocity social media feeds.
🏆 This feature launch provides Grok with a competitive advantage against competitors like ChatGPT, showcasing its unique potential.