%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
Welcome to the "AI Daily" section! 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 technical trends and learn about innovative AI product applications.
Fresh AI products Click to learn more:https://app.aibase.com/en
1. Meituan officially released the trillion-parameter open-source large model LongCat-2.0, which supports 1M ultra-long context natively
Meituan officially released and open-sourced the new generation of trillion-parameter large model LongCat-2.0. The model was trained and inferred on a domestic computing cluster, with a total of 1.6T parameters and native support for 1M ultra-long context. Its preview version ranked among the top three globally in monthly usage on the OpenRouter platform and performed outstandingly in multiple ecosystems. The release of LongCat-2.0 marks a major breakthrough in large-scale cluster training using domestic computing power. Its architecture is designed around real Agentic Coding tasks, introducing sparse attention mechanisms and zero-computation expert mechanisms, making the model perform exceptionally well in complex scenarios such as code understanding, mathematical reasoning, and long-range retrieval, even surpassing GPT-5.5 and Claude Opus4.6.
【AiBase Summary:】
🚀 Meituan released the trillion-parameter large model LongCat-2.0, supporting 1M ultra-long context.
🧠 Achieved distributed stability and efficient training through self-developed technology.
📈 Performed better than GPT-5.5 and Claude Opus4.6 in programming evaluations.
2. Say goodbye to "long text anxiety": Xiaohongshu RedKnot inference engine open-sourced, doubling the efficiency of long context processing
Xiaohongshu's technical team open-sourced the RedKnot inference engine, which significantly improved the efficiency of long text processing through an innovative KV Cache storage method, providing a new optimization path for AI inference systems.
【AiBase Summary:】
🧠 RedKnot improves long text processing efficiency by decomposing the way KV Cache is stored.
🚀 Test data shows that RedKnot significantly improves the speed of first character generation and single card concurrency.
📊 After open-sourcing, RedKnot provides a new idea for building lighter and more efficient AI inference systems.
3. Not just conversations! Douba App now has built-in map navigation, indicating AI large models are fully penetrating daily travel
Douba App launched its built-in navigation function, marking the acceleration of AI large model applications toward strong tool attributes, "super apps."
【AiBase Summary:】
📱 Douba App launched built-in navigation function, enhancing user experience.
🧭 Integrated Baidu Map technology to achieve native navigation process.
👥 Douba and Feishu have tried collaboration, but no social functions are planned.
4. Self-hosted AI gets a new convenience! Open source proxy OpenClaw's native iOS app is now available
As an open source AI assistant project, the release of OpenClaw's native iOS app marks an improvement in the mobile self-hosted AI agent experience, providing users with a smoother and more stable operation method.
【AiBase Summary:】
📱 OpenClaw's native iOS app is now available, offering users a more convenient self-hosted AI agent experience.
🔒 The app supports calling device capabilities while reminding users to pay attention to security configuration strategies.
🌐 Users can access various mainstream AI service APIs through OpenClaw and deeply integrate with local data.
5. Starting today, all free Gemini users in the U.S. can use personalized AI image generation features
The article introduces the personalized AI image generation feature launched by Google Gemini platform, which enables smarter image generation experiences by connecting user authorized data and provides privacy protection options.
【AiBase Summary:】
🧠 Personalized AI image generation features are now available to all free Gemini users in the U.S.
🎨 The system can connect user authorized data to intelligently generate images tailored to their interests.
🔒 Users can flexibly control the activation and access scope of personalized intelligent features.
6. Google AI Overview launches "Hot News Carousel" - Real-time breaking news directly embedded in AI summaries
Google introduced the "Hot News Carousel" feature in AI Overview on June 29, 2026, aiming to improve the efficiency of users obtaining real-time information and optimize the news recommendation experience.
【AiBase Summary:】
📱 Google introduced a "Hot News Carousel" in AI Overview, improving the efficiency of users obtaining real-time information.
📰 The new feature supports integration with users' preferred sources, optimizing the news recommendation experience.
🔄 Google has found a new balance between technological advancement and publishers' interests.
7. Zhiyuan Innovation releases Data Acquisition 2.0 Technology: Equipping robots with a smarter "brain"
Zhiyuan Innovation's embodied intelligence data acquisition 2.0 technology system has made breakthroughs in data operations, model evaluation, and scenario implementation, providing core driving force for robot development. This technology fills the industrial gap in the field of embodied intelligence in Chengdu and even Southwest China, and promotes the industry into a more systematic and refined development phase.
【AiBase Summary:】
🧠 The embodied intelligence data acquisition 2.0 technology system was released, improving data collection efficiency and flexibility.
📊 Built the first standardized integrated embodied intelligence model evaluation platform in the country.
🚀 Promoted the embodied intelligence industry in Chengdu and across the country into a new stage of systematic and refined development.
8. Huawei's openPangu 2.0 dual versions make their debut: Accelerating the bidirectional breakthrough of computing power and ecosystem in the Agent era
Huawei officially announced the open source of the 92 billion parameter openPangu-2.0-Flash model weights, basic inference code, and training and inference operators of the PanGu large model, aiming to accelerate the commercial innovation of artificial intelligence, build the intelligent infrastructure for the Agent era, and promote the Ascend developer ecosystem.
【AiBase Summary:】
🧠 Huawei's openPangu 2.0 dual versions made their debut, accelerating the bidirectional breakthrough of computing power and ecosystem in the Agent era.
🚀 The Flash version with 92B parameters has been open-sourced, balancing lightweight and high-concurrency inference performance.
📅 The openPangu 2.0-Pro model will be launched in July, with more components planned for the second half of the year.