AI Daily: Unitree Opensources UnifoLM-VLA-0 Large Model; Tencent Yuanbao's Internal Test Screenshots Leaked; Clawd Rebrands to OpenClaw

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/en

1. Ant Group Launches LingBot-VLA: Robotic Manipulation Enters the "Large Model Era"

Ant Group has released a visual-language-action (VLA) foundation model called LingBot-VLA, focusing on complex robotic manipulation in the real world. The model, trained on massive data, achieves general manipulation capabilities on different types of robots, marking an important advancement in embodied intelligence.

【AiBase Summary:】

🤖 LingBot-VLA uses an innovative "Hybrid Transformer" architecture, with Qwen2.5-VL as the multimodal backbone, capable of processing multi-view images and natural language instructions simultaneously.

🔍 Introduces the LingBot-Depth spatial perception model, enhancing 3D spatial reasoning through feature distillation, performing excellently in tasks like stacking and inserting.

🚀 The model has high data efficiency, requiring only about 80 demonstration data points for specific tasks to quickly adapt to new robots, and has open-sourced the entire training toolkit and model weights.

Details Link: https://arxiv.org/pdf/2601.18692

2. Yuque Opens UnifoLM-VLA-0 Large Model: Injecting "Physical Common Sense" into General Humanoid Robots

Yuque announced the open-source of the UnifoLM-VLA-0 large model, specifically designed for general humanoid robots, marking a key step forward in the field of embodied intelligence, where robot brains move from simple "text and image understanding" to having "physical common sense".

【AiBase Summary:】

🧠 Pre-training on robot operation data enables the model to understand the interaction rules of the physical world.

🌐 Deeply integrates text instructions with 2D/3D spatial details, enhancing spatial perception in complex environments.

⚙️ Integrates action chunk prediction and dynamics constraints, achieving unified modeling of long-term action sequences.

3. Another Name Change! Clawd Renamed OpenClaw GitHub Stars Exceed 100,000 and Spark Community Excitement

The article introduces the name change journey of the open-source personal AI assistant project from Clawd to Moltbot and finally to OpenClaw, along with its explosive growth on GitHub and core function updates. The project emphasizes open source, privacy protection, and security, and plans to enhance security and scalability in the future.

【AiBase Summary:】

🧩 The project has undergone multiple name changes and finally settled on OpenClaw, emphasizing open source and community-driven characteristics.

🚀 OpenClaw's GitHub star count has exceeded 100,000, featuring powerful functions and a focus on privacy protection.

🔒 The project continues to strengthen security measures and explore maintenance payment mechanisms to ensure the sustainability of the community.

Details Link: https://github.com/openclaw/openclaw

4. SenseNova-MARS Open-Sourced by SenseTime: Agentic VLM Empowers AI with "Independent Thinking and Action"

SenseTime officially open-sourced its first Agentic VLM model that supports dynamic visual reasoning and deep integration of image-text search—SenseNova-MARS. The model outperforms GPT-5.2 in multiple rankings, demonstrating strong performance and autonomous planning capabilities.

【AiBase Summary:】

🧠 SenseNova-MARS is an Agentic VLM model that supports dynamic visual reasoning and deep integration of image-text search.

📊 SenseNova-MARS performs well in evaluations such as MMSearch and HR-MMSearch, surpassing GPT-5.2.

🛠️ Through dual-phase training, SenseNova-MARS grants AI the ability for autonomous planning and handling complex tasks.

5. Tencent's "Yuanbao Pai" Internal Test Screenshot Leaked, Social + AI Rebuilds Entertainment Experience

Tencent's AI assistant 'Yuanbao' caused attention due to the leak of internal test screenshots of the social function 'Yuanbao Pai'. This function added features like 'Listen to Music Together' and 'Watch Video Together', integrating the resources of QQ Music and Tencent Video, supporting synchronized viewing and listening with friends and real-time interaction. In addition, 'Yuanbao Pai' achieved cross-platform social integration between WeChat and QQ contacts, breaking down the barriers between the two social ecosystems for the first time. AI plays an active role in group chats, providing intelligent summaries, task monitoring, and other tool-based services, and can also generate funny pictures and adjust the atmosphere. Ma Huateng plans to invest 1 billion yuan in red packet resources to promote this function, aiming to recreate a social miracle during the Spring Festival.

【AiBase Summary:】

🎧 Integrates QQ Music and Tencent Video resources, supporting synchronized watching and listening

🤝 Achieves cross-platform social integration between WeChat and QQ contacts

🤖 AI in group chats provides intelligent summaries, task monitoring, and cold-breaking services

6. Musk Invests Heavily in AI Blueprint: xAI Plans to Launch Customized Generative Games and Films Next Year

Musk is confident in xAI's vision, planning to launch real-time generated, high-quality, and highly personalized films and video games next year. Although there are already some AI-assisted game developments in the market, Musk aims to achieve full AI production from gameplay, art design to level generation, which has attracted widespread industry attention and skepticism.

【AiBase Summary:】

🎮 Musk predicts that xAI will release AI-generated games in 2025, challenging traditional content creation paradigms.

🖼️ He emphasizes that game and film content will achieve real-time generation, high quality, and high personalization.

🤔 There is currently no impressive AI-generated game demo, and whether the goal can be achieved remains to be seen.

7. Claiming $3 Billion! Anthropic Sued by Music Publishers for Mass Copyright Infringement

The article reports that music publishers sued Anthropic, accusing it of mass copyright infringement during the training of AI models, and demanding $3 billion in compensation.

【AiBase Summary:】

🎼 Anthropic is accused of mass copyright infringement involving over 20,000 copyrighted songs.

💰 The claim may exceed $3 billion, becoming one of the largest non-collective copyright lawsuits in U.S. history.

⚖️ Publishers accuse Anthropic's business practices of being inconsistent with its claimed compliance and ethical image.

8. Competing for Cloud Dominance: ByteDance and Alibaba Plan to Launch New Generation Large Models During the Spring Festival

The article explores the competition between ByteDance and Alibaba regarding the release of new generation underlying model technologies around the Spring Festival, analyzing the strategies both sides have taken in their struggle for domestic technological leadership and future cloud service market dominance, as well as the impact on the industry's future prospects.

【AiBase Summary:】

🧠 ByteDance plans to launch three core models, covering large-scale language processing, image generation, and video generation fields.

💻 Alibaba will launch the flagship model Qwen3.5, focusing on improving logical reasoning, mathematical calculation, and code writing capabilities.

💰 The Chinese cloud market is expected to reach nearly $90 billion by 2030, and the technology competition will reshape the future landscape of the tech industry.