Welcome to the "AI Daily" column! This is your guide to exploring the world of artificial intelligence every day. Each day, we present you with the latest content in the AI field, focusing on developers to help you understand technology trends and innovative AI product applications.
Fresh AI products click to learn more:https://top.aibase.com/
1. Kunlun Wanyi officially launched SkyReels-A3 model: photos can mouth along with voice
The SkyReels-A3 model developed by Kunlun Wanyi Group is based on the DiT video diffusion model, achieving audio-driven digital human creation. This model enables people in static images or videos to speak or sing according to the content of the voice, and supports changing dialogues and camera movement control functions, providing an efficient and convenient AI technical solution for advertising, live streaming sales, and music MVs.
【AiBase summary:】
📷 SkyReels-A3 can dynamically perform characters in static images or videos according to the voice content.
🎥 Supports single shot video output up to 60 seconds, multi-shot support for unlimited duration, meeting different creative needs.
🔄 Provides 8 preset camera movement parameters, intensity adjustable, achieving professional-level camera effects.
Details link: https://skyworkai.github.io/skyreels-a3.github.io/
2. Elon Musk's xAI announces permanent free access to Grok 4 AI model
xAI company announced that the Grok4 artificial intelligence model will be permanently open for free to global users.
【AiBase summary:】
🤖 Grok4 artificial intelligence model will be permanently open for free to global users.
⚙️ Provides Auto mode and Expert mode, meeting different user needs.
🌐 Free access may promote the popularization and application of AI technology.
3. Open AI releases GPT-5 prompt guide: unlocking new horizons in AI programming and multimodal interaction
The article introduces in detail the GPT-5 model released by Open AI and its official prompt guide, emphasizing its improvements in complex tasks, programming, and multimodal interaction. The guide provides optimization strategies, such as adjusting the degree of reasoning, controlling the tendency of agent behavior, and using tool prefaces, to help users maximize the potential of GPT-5.
【AiBase summary:】
🧠 GPT-5 improves the performance of agent tasks, code generation, and instruction following through precise prompt design.
💻 Supports generating front-end interfaces, debugging large codebases, and improving code generation efficiency with the Responses API.
🖼️ Introduces multimodal interaction features, including text, image, and voice processing, as well as personalized settings, enhancing practicality.
Details link: https://cookbook.openai.com/examples/gpt-5/gpt-5_prompting_guide
4. Baidu Search fully launches AI search function on PC
Baidu Search has fully launched a series of AI functions on the PC end, transforming traditional information entry points into task centers. The newly added "Super Intelligent Double-line Box" and "Workbench" modules integrate AI reading, AI writing, and AI PPT tools, improving user search efficiency and office experience. At the same time, the monthly active users of Baidu AI Search have exceeded 322 million, firmly ranking first in the domestic AI search industry.
【AiBase summary:】
🧠 Baidu Search PC end has fully launched AI functions, improving user search experience.
🛠️ New "Workbench" module integrates AI reading, writing, and PPT tools.
📈 Monthly active users reached 322 million, Baidu firmly ranks first in the domestic AI search industry.
5. Windows 11 Copilot app freely accesses GPT-5, with usage limits much lower than ChatGPT
Microsoft announced that the Copilot app in Windows 11 and Windows 10 has fully supported the GPT-5 intelligent mode. This feature is implemented through web routing technology, allowing users to enable the intelligent mode without updating, and the usage limits are more relaxed than those of ChatGPT.
【AiBase summary:】
🌟 Copilot now supports GPT-5 intelligent mode, offering a smoother user experience.
💬 Compared to ChatGPT, Copilot has more relaxed usage limits, increasing freedom.
🖥️ Users can access Copilot and GPT-5 for free through simple steps, making it easy to obtain information.
6. Surpassing OpenAI! Baichuan Intelligence's open-source medical large model Baichuan-M2 leads globally
Baichuan-M2, the open-source medical-enhanced large model released by Baichuan Intelligence, scored 60.1 on the HealthBench evaluation, surpassing OpenAI's gpt-oss120b model and leading other open-source large models internationally. The model has been optimized for extreme lightness, enabling deployment on a single card, significantly reducing costs for medical institutions. Meanwhile, Baichuan-M2 demonstrates capabilities comparable to GPT-5 in handling complex medical issues, showcasing strong application potential.
【AiBase summary:】
🌟 Baichuan-M2 scored 60.1 on the HealthBench evaluation, becoming a globally leading open-source medical model.
💡 The model has been optimized for lightness, enabling deployment on a single card, significantly reducing costs for medical institutions.
🚀 Baichuan-M2 demonstrates capabilities comparable to GPT-5 in handling complex medical issues, showing great application potential.
Details link: https://huggingface.co/baichuan-inc/Baichuan-M2-32B
7. Apple announces GPT5 will be integrated into iOS 26: iOS 26 will integrate ChatGPT5
Apple announced that the ChatGPT-5 model will be integrated into the upcoming iOS 26 system, which will significantly enhance the performance of Apple Intelligence and bring a series of new features, such as real-time translation and content search optimization. Users do not need an OpenAI account to use these features, but associated accounts can enjoy more benefits.
【AiBase summary:】
🧠 ChatGPT-5 will be integrated into iOS 26, enhancing the performance of Apple Intelligence.
🌐 New real-time translation features improve cross-language communication experiences.
💰 Associated OpenAI accounts can enjoy subscription discounts, providing more choices.
8. Google Launches BlenderFusion: A New Framework for 3D Visual Editing and Generation Synthesis
Google's BlenderFusion is an innovative framework designed to enhance 3D visual editing and generation synthesis capabilities, providing designers and creators with more intuitive and efficient creative tools.
【AiBase summary:】
🎨 BlenderFusion integrates advanced 3D editing tools and diffusion models, achieving efficient 3D visual editing and generation synthesis.
🛠️ The framework's workflow includes three stages: layering, editing, and synthesis, allowing users to easily edit 3D objects and generate final images.
📈 Google's BlenderFusion enhances the ability to handle complex scenes through model optimization, helping designers realize their creativity.
Details link: https://blenderfusion.github.io/
9. Ultra-small TTS model Kitten TTS: parameter count only 15 million
Kitten TTS is an open-source lightweight text-to-speech model with only 15 million parameters and a volume less than 25MB, suitable for deployment on various devices. It supports running without a GPU and can achieve high-quality speech synthesis on a regular CPU, and provides simple installation and usage guides, allowing users to quickly get started.
【AiBase summary:】
🐱 Kitten TTS is an open-source lightweight text-to-speech model, with a volume less than 25MB, suitable for various devices.
⚡ The model supports running without a GPU, ensuring high-quality speech synthesis on regular CPUs.
🚀 Kitten TTS has provided simple installation and usage guides, allowing users to quickly start and generate audio.
Details link: https://huggingface.co/KittenML/kitten-tts-nano-0.1
10. MiniCPM-V 4.0, a small and powerful vision model, runs more smoothly on mobile devices
MiniCPM-V 4.0, the latest version of the MiniCPM-V series, performs excellently in visual understanding, multi-image, and video processing, and achieved a high score of 69.0 in the OpenCompass evaluation, surpassing multiple similar models. It is specifically designed for mobile devices, with fast response speed and no overheating issues, and provides various usage methods and open-source tools, making it easy for users to get started.
【AiBase summary:】
🌟 MiniCPM-V4.0 scored 69.0 in the OpenCompass evaluation, surpassing multiple similar models.
📱 The model is specifically designed for mobile devices, with fast response and no overheating issues.
📚 Open-source iOS apps and detailed usage guides make it easier for users to get started.
Details link: https://huggingface.co/openbmb/MiniCPM-V-4
11. Stripe report: AI economy is growing rapidly, revenue growth speed exceeds SaaS by three times
Stripe's latest analysis report reveals the rapid development of the AI economy, including revenue growth speed, global market expansion, and business model innovation. The report points out that AI startups achieve revenue milestones at a speed far exceeding previous tech companies and possess the 'innate globalization' gene.
【AiBase summary:】
🚀 AI companies achieve revenue growth speed far exceeding traditional SaaS companies, reaching $1 million annual revenue in just 11.5 months.
🌍 AI companies have the globalized gene from the beginning, covering twice as many countries in the first year as SaaS companies.
💡 Business models continue to innovate, usage-based billing and results-based billing models are increasingly popular, driving AI companies to monetize quickly.