Welcome to the AI Daily column! Your daily guide to exploring the world of artificial intelligence. We bring you the hottest AI news, focusing on developers and helping you understand technology trends and innovative AI product applications.
New AI Products Learn More: https://top.aibase.com/
1. Robin Li Unveils Wenxin's Dual Star Large Models: X1Turbo Targets DeepSeek, 4.5Turbo Surpasses GPT-4o
At Baidu's Create Developer Conference, Robin Li launched the new generation Wenxin large model X1Turbo, which boasts significant advantages in both performance and price. X1Turbo's input and output prices are 1 yuan and 4 yuan respectively, only 25% of its competitor DeepSeek-R1. Furthermore, Wenxin 4.5Turbo has even lower input and output prices of 0.8 yuan and 3.2 yuan respectively, and excels in various benchmark tests, surpassing GPT-4o. The release of these two models will further intensify competition in China's AI large model market, especially in price-sensitive and performance-driven application scenarios.
【AiBase Summary:】
💡 Wenxin large model X1Turbo offers significantly improved performance with an input price of just 1 yuan and an output price of 4 yuan, making it highly competitive.
📊 Wenxin 4.5Turbo is faster and 80% cheaper, with input and output prices of 0.8 yuan and 3.2 yuan respectively.
🏆 Wenxin 4.5Turbo achieved an average score of 77.68 in benchmark tests, surpassing GPT-4o's 72.76, demonstrating excellent performance.
2. Baidu Launches AI Open Program to Empower Developers to Embrace MCP
At Baidu's Create Developer Conference, Robin Li announced a series of initiatives supporting AI applications, launching the "AI Open Program" to provide comprehensive support for developers. This program utilizes diverse content and service distribution mechanisms to meet user demands for AI services while generating traffic and revenue for developers. Li emphasized Baidu's commitment to lowering the barrier to entry for developers, fostering rapid innovation in AI applications, and plans to cultivate 10 million AI talents in the next five years to embrace the intelligent new era.
【AiBase Summary:】
🚀 Baidu's "AI Open Program" provides comprehensive support for developers, promoting the development of AI applications.
💡 The program integrates various innovative applications, meeting user needs for the latest AI services and enhancing the revenue potential for developers.
🏆 The "Wenxin Cup" entrepreneurship competition has been launched, with investments of up to 70 million yuan, aiming to cultivate 10 million AI talents.
3. OpenAI Makes Lightweight Version of Deep Research Free, Powered by o4-mini
OpenAI announced the free release of a lightweight version of its AI research tool, Deep Research, marking a significant step towards the popularization of AI technology. Deep Research can independently complete complex research tasks and generate detailed research reports, a benefit now extended to free users. The lightweight version is powered by the o4-mini model; while reports are shorter, it retains core intelligence and analytical capabilities. This move not only expands the user base but also addresses market competition, further solidifying ChatGPT's market position.
【AiBase Summary:】
🧠 Deep Research is OpenAI's AI research agent capable of independently completing complex research tasks and generating detailed reports.
📈 The lightweight Deep Research is powered by the o4-mini model; while reports are shorter, it retains core intelligence and in-depth analytical capabilities.
🌍 Free user access to Deep Research is in the testing phase. OpenAI promises to share more details soon to meet user needs.
4. Jidream Video 3.0 Internal Testing: Smooth Camera Work, Precise Capture of Facial Expressions
Internal testing of Jidream Video 3.0 showcases significant advancements in video creation, particularly in smooth camera work and capturing human emotions. The new model handles diverse scenes and supports high-definition quality, demonstrating enhanced artistic expression. Although still in the testing phase, its powerful features and precise performance suggest limitless possibilities for future AI video creation, making it highly anticipated by creators.
【AiBase Summary:】
🎬 Rich camera language, supporting various professional camera techniques, enhances video storytelling and visual impact.
🎨 Supports diverse styles, including surrealism, cartoons, and nature documentaries, meeting creators' artistic visions.
🐾 Unique animal expressiveness; the model can give animal characters vivid movements and personalities, enhancing overall performance.
5. Baidu Launches Content Operating System "Cangzhou OS," Baidu Wenku's AI Monthly Active Users Near 100 Million
At the Baidu Create Conference on April 25th, Robin Li launched the world's first content operating system, "Cangzhou OS," aiming to enhance the intelligence and efficiency of content management. The core component "Chatfile Plus" can perform in-depth analysis of multimodal content, while "AI Notes," jointly launched by Baidu Wenku and Baidu Netdisk, provides users with convenient learning and content organization tools. With the popularization of AI technology and continuous user experience improvements, Baidu will continue to increase its investment in AI to meet the growing needs of modern users.
【AiBase Summary:】
🌟 Baidu launches the world's first content operating system, "Cangzhou OS".
📈 Baidu Wenku and Baidu Netdisk's AI monthly active users have reached nearly 100 million.
📝 The newly launched "AI Notes" is the industry's only multimodal AI note-taking tool.
6. Baidu Wenku and Baidu Netdisk Jointly Release GenFlow and AI Notes
At the Create 2025 Baidu AI Developer Conference on April 25th, Baidu Wenku and Baidu Netdisk launched two innovative AI tools: "GenFlow" and "AI Notes." These products aim to improve user work and learning efficiency, leveraging large model technology across multiple scenarios. GenFlow automatically plans tasks and generates high-quality content through simple instructions, while AI Notes seamlessly connects video learning with note-taking, automatically generating structured multimodal notes. These tools not only enhance user productivity but also distinguish Baidu Wenku and Netdisk in the AI era.
【AiBase Summary:】
📈 GenFlow quickly generates high-quality content; users can automatically plan tasks with simple instructions.
🎓 AI Notes seamlessly integrates video learning and note-taking, automatically generating structured multimodal notes.
🌐 Baidu Wenku and Netdisk's jointly launched AI tools cover multiple scenarios, serving a total of 1 billion users and boosting productivity.
7. Pixverse Launches MCP: One-Click Access to a New Realm of AI Video Generation
With the rapid development of generative AI technology, Pixverse's Model Context Protocol (MCP) has revolutionized video creation. MCP allows users to generate high-quality videos using natural language prompts, eliminating the need for complex development environments and significantly lowering the technical barrier. Its openness and flexibility empower content creators, marketers, and developers to create more freely, while also providing new opportunities for the developer community. This innovation enhances user experience and promotes the popularization of AI video generation.
【AiBase Summary:】
🚀 MCP is a protocol designed for AI video generation, allowing users to generate videos via natural language prompts.
💻 The protocol supports multi-resolution output and diverse scene descriptions, improving the structured nature of video content.
📈 MCP's openness provides developers with opportunities for customization and extension, promoting the popularization of AI video creation.
8. Tavus Releases SOTA Lip-Synchronization Model Hummingbird-0: Revolutionizing Zero-Shot Lip-Synchronization Technology
Tavus's recently released Hummingbird-0 model has achieved a breakthrough in lip-synchronization technology, marking a new era of zero-shot lip synchronization. This model not only boasts high-precision lip synchronization but also surpasses other models on the market in visual quality and identity preservation. Hummingbird-0 has wide-ranging applications, including content creation and multilingual dubbing, significantly improving video editing efficiency and quality.
【AiBase Summary:】
🚀 Hummingbird-0 is the current state-of-the-art zero-shot lip-synchronization model, achieving high-precision synchronization without model training.
🌍 The model is suitable for various applications, including user-generated content, dubbing, and personalized videos, reducing editing time costs.
🏆 Tavus's comparative tests demonstrate that Hummingbird-0 surpasses other industry-leading tools in visual quality and synchronization accuracy.
9. Doubao 1.5 Deep Thinking Model Launches on Edge Large Model Gateway; Millions of Tokens Available for Free
ByteDance's Volcano Engine's Doubao 1.5 Deep Thinking model is now available on the edge large model gateway, providing users with up to 5 million tokens for free. This high-performance AI model excels in inference and creative writing, supporting multimodal inference and significantly improving the usability and efficiency of AI services. Through edge computing, users can access various large models quickly and reliably, promoting the widespread application of AI technology.
【AiBase Summary:】
🚀 Doubao 1.5 Deep Thinking model offers up to 5 million tokens for free, supporting various use cases.
💡 The model uses MoE architecture, with significant parameter optimization, offering high concurrency and low latency.
🌐 The edge large model gateway is compatible with over 100 mainstream large models, improving the speed and reliability of AI services.
10. Adobe's New Firefly Platform Integrates OpenAI and Google AI Models, Upgrading Creative Tools
Adobe's launch of the new AI model suite Firefly marks a significant advancement in creative design. Firefly integrates advanced technologies from multiple partners to enhance user creativity within Creative Cloud. Generative AI allows users to quickly generate creative content, saving time. Firefly's easy integration allows even creative professionals without programming backgrounds to easily use these powerful tools. In the future, Firefly will have a profound impact on the design industry.
【AiBase Summary:】
✨ The Firefly platform integrates advanced AI technologies from OpenAI and Google, improving creative efficiency.
🖼️ Users can quickly generate related images or designs from simple text descriptions, saving creation time.
🔧 Firefly's integration is convenient, allowing creative professionals to easily get started without a programming background.
11. ImageSlider 2.0 to Join Core Product Line, Image Generation Capabilities Significantly Upgraded
The Gradio team is about to launch ImageSlider 2.0 as part of its core product line, bringing a range of new features and performance enhancements. This update aims to improve user experience, expand creative options, and increase generation efficiency. The new version supports multiple layouts and high-resolution image generation, suitable for e-commerce, digital art, and other fields. The community response has been enthusiastic, with users already experiencing its commercial potential in testing.
【AiBase Summary:】
✨ Enhanced image sliding experience, supporting dynamic transitions and interactive navigation, optimizing mobile and desktop user experiences.
🎨 Provides diverse layout options; users can customize image arrangement according to their needs, suitable for various display scenarios.
🚀 Supports high-resolution image generation and video playback, enhancing the display effects of e-commerce and digital art.
Details Link: https://github.com/gradio-app/gradio/pull/11027
12. Robin Li Discusses DeepSeek's Existing Pain Points, Calling DeepSeek Slow and Expensive
At today's Create 2025 AI Developer Conference, Baidu founder Robin Li detailed the application status and challenges of the DeepSeek model. He pointed out that although DeepSeek has made progress in intelligent customer service and search enhancement, it still has technical limitations, such as the inability to handle multimodal content and slow response speeds. Li emphasized that future AI models need multimodal capabilities, and reducing costs is key to promoting the popularization of AI applications. Baidu's new version of the Wenxin large model aims to address these issues to better serve enterprise clients.
【AiBase Summary:】
🛠️ The DeepSeek model currently only supports text processing and cannot generate multimodal content, limiting its application in high-risk areas.
💰 Wenxin 4.5Turbo and X1Turbo versions optimize performance and cost, aiming to lower the deployment barrier for enterprises.
📈 Baidu strives to balance model capabilities and commercialization through technological iteration and cost restructuring.
