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 technology trends and learn about innovative AI product applications.
Fresh AI products Click to learn more:https://app.aibase.com/zh
1. Doubao Mobile Assistant Consumer Edition Released: AI Button with Fingerprint Activation, Cross-App Agent Upgraded Again
The consumer version of Doubao Mobile Assistant has been fully upgraded in terms of interaction, memory, agent execution, and security, pushing mobile AI from "Q&A" to "automated tasks."
【AiBase Highlights:】
📱 AI button supports fingerprint authentication, wakes up with one touch on the lock screen
🧠 Improved memory capabilities, can securely retrieve local data
🔒 Enhanced security mechanisms, using a combination of endpoint and cloud processing for data
2. Suno Launches v6 Music Model: Collaborates with Warner, BMG, Create with Natural Language, Edit, Mix, Sample
Suno launches the new generation music model v6 in collaboration with music companies such as Warner and BMG. The model comes in the form of a model suite, including a flagship version, an exploration version, and a lightweight version, meeting the needs of different creators. v6 supports multi-modal input and can precisely modify song content through natural language, marking a deep integration of AI and the music industry.
【AiBase Highlights:】
🚀 Suno v6 music model developed in collaboration with music companies like Warner and BMG is an important step towards making music creation accessible to more people.
🎵 v6 is released as a model suite, including a flagship version, an exploration version, and a lightweight version, to meet the needs of different creators.
🖼️ Supports multi-modal input (text, audio, images, video), and can precisely modify song content through natural language.
3. ElevenLabs Releases Music v2.5 Audio Model, Also Opens APP and API Access
ElevenLabs launched a new version of ElevenMusic Music v2.5 and opened access to global developers and users via mobile applications and APIs. This version performs well in genres such as R&B, soul, hip-hop, rock, and orchestral music, and offers free and professional versions. In addition, ElevenLabs emphasized that Music v2.5 is trained on a legally authorized music dataset to avoid copyright disputes.
【AiBase Highlights:】
🌟 ElevenLabs launches Music v2.5, improving audio quality and naturalness.
💡 Offers free and professional versions to meet the needs of different users.
🔒 Emphasizes copyright compliance to avoid copyright disputes with competitors like Suno.
4. Unitree Humanoid Robot G1+ Released: Motor, Vision, Interaction, and Battery Life Fully Enhanced
Unitree Technology released the humanoid robot G1+, which has made multiple upgrades in motion performance, perception interaction, and smart experience, including neck flexibility, motor output and thermal management, visual and tactile perception, power supply and battery life, and long-range voice interaction capabilities. The standard version costs 95,000 yuan, and the G1+ EDU version for research and in-depth development is also available.
【AiBase Highlights:】
🧠 The humanoid robot G1+ adds a +2DOF neck structure, enhancing flexibility and motion performance.
📷 Equipped with dual vision systems, 3D LiDAR, and a wide-angle camera on the abdomen, it enhances depth and spatial understanding capabilities.
🔋 The G1+ adds an external power port and quick-release smart battery, with a battery life of about 2 hours, meeting the needs of long-duration online scenarios.
5. Zhipu Secures $5 Billion New Funding, Accelerating Next-Generation GLM Recursive Self-Improvement
Zhipu announced the completion of approximately $5 billion in funding, primarily used for the development of the next-generation GLM foundation model and the construction of a self-training system, while advancing the adaptation of domestic chips and optimizing inference efficiency to achieve efficient technological iteration and expansion.
【AiBase Highlights:】
🧠 The training environment for the next-generation GLM model is built on the intelligence of the previous generation GLM, achieving a recursive self-improvement cycle.
⚙️ Funds will be used for automated generation and screening of training data, building task environments, and improving the model's long-range reasoning ability.
💻 Engineering efforts will focus on domestic chip adaptation, deep operator development, and inference efficiency optimization to maximize output under hardware conditions.
6. New Apple TV May Be Launched Quietly This Autumn: Upgraded Chip, Siri Becomes Core Upgrade
Apple plans to quietly launch a new Apple TV box this autumn, with core upgrades including integration into the Apple Intelligence ecosystem, stronger chips, and enhanced capabilities of the Siri AI assistant, while strengthening its role as a smart home hub.
【AiBase Highlights:】
🍎 The new Apple TV box will fully integrate into the Apple Intelligence ecosystem and be equipped with a stronger chip to improve performance.
🎙️ The Siri AI assistant will achieve deep situational understanding, supporting complex natural language commands and cross-platform recommendations.
🏠 The new Apple TV will strengthen its position as a smart home hub, supporting Thread and Matter protocols to optimize the experience of family device coordination.
7. JD.com Unveils Interactive Audio-Visual World Model at JDD: EchoWM Open Source, Long Video Generation Reaches Over 10 Minutes
JD.com unveiled JoyAI-Echo1.5 and EchoWM at JDD, achieving ultra-long audio-video generation and interactive audio-visual world models, while showcasing the JoyAI basic model matrix, promoting AI's layout from the digital world to the physical world.
【AiBase Highlights:】
🎥 JoyAI-Echo1.5 achieves ultra-long audio-video generation, supporting continuous output of over 10 minutes.
🌐 EchoWM open source and achieves interactive audio-visual world model, supporting multi-round exploration and synchronized sound changes.
📦 JD.com showcases the JoyAI basic model matrix, covering image and video generation, multimodal understanding, and other capabilities.
8. DeepSeek-V4.1-Flash Launched on Qwen AI Platform, API and Token Plan Also Opened
The launch of DeepSeek-V4.1-Flash brings new technical breakthroughs to the Qwen AI platform, with excellent performance and cost control, suitable for enterprise tasks such as long document parsing and customer service knowledge base.
【AiBase Highlights:】
🧠 DeepSeek-V4.1-Flash uses MoE architecture, with input activation of about 8B and output activation of about 16B, supporting text and image understanding.
💰 Significant cost optimization, KV Cache reduces HBM demand to 1/4 of the previous generation, and SSD storage demand is reduced to 1/8.
🌐 Supports multi-region calls, covering Beijing, Singapore, and Global regions such as the US, Germany, Japan, and Hong Kong.
