Welcome to the [AI Daily] column! This is your daily guide to exploring the world of artificial intelligence. Every day we bring you hot topics in the AI field, focusing on developers and helping you stay on top of technology trends and innovative AI product applications.
Fresh AI ProductsClick to learn more：https://app.aibase.com/zh
1. Kling Kling4.0 fully launched, with a maximum single video of 30 seconds and support for continuation shooting
Kuaishou's Kling AI has officially launched Kling4.0, with further upgrades in video realism, camera movement, lip sync and audio performance. It can generate videos up to 30 seconds in a single pass, and multiple continuations can extend the duration to 2 minutes.
[AiBase Highlights:]
🎥 Maximum single video of 30 seconds; multiple continuations can reach 2 minutes
🎙️ Upgraded lip matching, stereo sound and high-definition HDR image quality
🧩 Supports multi-image and multi-video input, and integrates MCP and CLI
2. Claude Opus5.5 Turbo mode fully opened, generation speed boosted by 2.5x
Anthropic has opened the Claude Opus5.5 Turbo mode to all developers, increasing output speed to 2.5 times that of the standard mode without weakening the model's reasoning ability, and optimizing API call cost management.
[AiBase Highlights:]
⚡ Turbo mode fully opened; no more waiting in queue to apply
🧠 Retains original model capabilities, with output speed increased by 2.5x
💰 New usage monitoring options help developers control costs
3. Jia Yueting launches another new product: mini humanoid robot priced at $2,490
Jia Yueting's Faraday Future has released the mini humanoid robot FF Nova, priced at $2,490, targeting the developer and research education markets, with support for bipedal walking, voice interaction and open development.
[AiBase Highlights:]
🤖 FF Nova priced at $2,490, positioned as an entry-level humanoid robot
🦿 Supports bipedal walking, autonomous navigation and voice interaction
🛠️ Open SDK and ROS2 make secondary development easy for developers
4. Meta open-sources super-large model Muse Spark Open, and plans to make it run locally on Mac
Meta has released the open-weight large model Muse Spark Open and partnered with Apple to optimize local deployment on Mac, while supporting multiple chips and inference platforms, but running it in full requires expensive high-memory hardware.
[AiBase Highlights:]
🧠 The model has a total of 816 billion parameters and supports million-level context
🍎 Partnering with Apple on optimization and adaptation to advance local large model operation
💻 Supports multiple chip platforms, though top-spec hardware is costly
5. Xiaomi open-sources spatial audio large model, enabling AI to not only hear but also tell direction
Xiaomi has fully open-sourced MiDashengLMSpatial, a large spatial audio understanding model that adds spatial perception and deep reasoning capabilities on top of traditional audio understanding; in the future it will be applied to smart homes and vehicle cockpits.
[AiBase Summary:]
🔊 The first open-source end-to-end spatial audio understanding large model
🧠 Combining sound recognition, spatial awareness and deep reasoning
🚗 Planned integration with Xiaomi vehicles and the smart home ecosystem
6. China Literature launches premium AI animation and upgrades its platform to help creators make content in one stop
China Literature Group has released its first batch of premium AI animation projects and upgraded DramaBuddy into OPC, a one-stop creation platform that integrates IP resources, AI production, multilingual visual transfer and global distribution, helping individual creators produce content efficiently.
[AiBase Summary:]
🎬 Released the first batch of premium AI animation projects including "Cyber Hero Chronicles"
🛠️ Upgraded the OPC platform, covering the full workflow from IP integration to AI production
🌍 Opened up multilingual overseas channels and global distribution routes
7. OpenAI upgrades the ChatGPT browser extension; AI can now get things done across web pages for you
OpenAI has upgraded the ChatGPT browser extension, enabling the AI to read content from multiple web pages, organize materials, fill in forms and draft emails, and even execute browser tasks directly after authorization.
[AiBase Summary:]
🌐 Supports reading multiple web pages and comparing information across pages
🤖 The AI can fill in forms, organize materials and draft emails
🔐 Web actions require authorization; OpenAI Atlas is officially retired
8. OpenAI announces that several old models will soon be retired; developers should hurry to migrate
OpenAI has announced that GPT5.5 is being withdrawn from ChatGPT, ChatGPT Work and signed-in Codex, while the relevant APIs are temporarily unaffected; several legacy API model versions, including GPT4o, are scheduled to be deactivated on January 14, 2027.
[AiBase Summary:]
📢 GPT5.5 has already been withdrawn from ChatGPT and some other products
⏳ Several legacy API models will be deactivated on January 14, 2027
🔧 Developers should check their model configurations and complete migration in advance