AI Daily: Tmall Launches AI Space Station; Gemini 3.8 Flash Released; Baidu's Basic Model Department Undergoes Major Changes

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、Tmall officially launched AI Space Station: Alibaba Cloud, Zhipu, Kimi, MiniMax join together

Tmall officially launched the AI Space Station, integrating resources from leading large model manufacturers such as Alibaba Cloud, Zhipu, Kimi, and MiniMax, providing users with convenient AI service subscription and payment methods, while expanding the AI smart hardware market and promoting AI commercialization to a broader consumer market.

【AiBase Summary:】

🛒 The AI Space Station has been launched, with major large model manufacturers joining the e-commerce ecosystem

💻 Tmall offers various AI service subscription plans to lower the user's usage barrier

📈 E-commerce expands into AI smart hardware, creating an integrated software and hardware consumer experience

2、Gemini 3.8 Flash released, six weeks of continuous updates but not as powerful as expected

Google released three models in the Flash series within six weeks, among which Gemini 3.8 Flash was highly anticipated, but it showed a gap compared to official demonstrations in actual tests. Although it performed well in various benchmark tests, it still had issues with superficial understanding and stability when dealing with complex tasks.

【AiBase Summary:】

🔥 Gemini 3.8 Flash performs well in multiple benchmark tests, approaching top models like Claude Opus5.

💡 This model improves its ability to handle complex tasks through more reasoning steps and tool calls, but may consume more tokens.

📉 Actual testing shows that it lags behind official examples in details, and its stability still needs improvement.

3、Baidu's Basic Model Department undergoes major changes: North American large model experts and former DeepSeek core members join, enhancing the Wenshi multimodal model

Baidu's Basic Model Research Department (BMU) recently made important talent adjustments by introducing Wu Qin, a senior researcher from the North American Frontier Lab, and Wei Haoran, a former researcher from DeepSeek, to comprehensively strengthen the pre-training capabilities and multimodal algorithm research of the Wenshi large model. These measures come at a critical time for BMU head Sun Tianxiang, aiming to keep up with the development pace of top international AI laboratories.

【AiBase Summary:】

🧠 Introducing Wu Qin, a senior researcher from North American Frontier Lab, to enhance the pre-training capabilities of the Wenshi large model

👨‍💻 Former DeepSeek researcher Wei Haoran joins BMU, serving as the head of the Wenshi multimodal algorithm team

🚀 Baidu's Basic Model Research Department accelerates organizational structure and core technology upgrades

4、Google Launches Generative AI Image Tool Google Pics, Competing with Canva and Adobe Express

Google officially launched the generative AI image tool Google Pics, aiming to reduce the threshold for producing high-quality visual content and promote the intelligent transformation of enterprise productivity tools.

【AiBase Summary:】

🖼️ Google Pics is Google's image generation and editing tool, targeting enterprise customers and advanced users.

🧠 This tool is based on Google's self-developed Nano Banana model, using a natural language prompt-driven design logic.

👥 Supports multi-version selection, team collaboration, and integration with Google Docs and Google Slides.

5、Doubao Work Updates: Parallel Agents Improve Efficiency, Mac Version "Operate Computer" Lets AI Take Over the Desktop

Doubao Work introduced the "Parallel Agents" and "Mac Version 'Operate Computer'" features this week, marking the transformation of AI assistants from conversation tools to all-around agents. These features improve task processing efficiency and expand the application boundaries of AI, providing users with a smarter desktop assistant experience.

【AiBase Summary:】

🧠 The parallel agents feature improves task processing efficiency, enabling collaborative work and delivering multiple forms of results.

💻 The "Operate Computer" feature supports the Mac system, allowing local desktop operations via GUI, enhancing the AI-user collaboration experience.

🔄 Doubao Work continues to update, releasing new features every Wednesday, optimizing AI's desktop application capabilities.

6、Valuation Approaching $5 Billion: Moonlight Dark Face Secretly Files for Hong Kong IPO, AI Unicorn Accelerates Its Embrace of the Capital Market

Moonlight Dark Face is accelerating its capitalization process, having submitted A1 listing application documents to the Hong Kong Stock Exchange and planning to raise funds at a pre-money valuation of $5 billion, which may be its last round of financing before the IPO. Previously, the company completed a $3.5 billion F-round financing, with a post-money valuation reaching $3.5 billion, and the market showed strong interest.

【AiBase Summary:】

🚀 Moonlight Dark Face has submitted A1 listing application documents to the Hong Kong Stock Exchange, starting the HK IPO process.

💰 Moonlight Dark Face is currently raising funds at a pre-money valuation of $5 billion, possibly the last round before the IPO.

📈 In July, Moonlight Dark Face completed a $3.5 billion F-round financing, with a post-money valuation of $3.5 billion, and the market showed strong interest.

7、Tencent Launched Public Welfare Industry Version of WorkBuddy, 15 Intelligent Agents Cover 42 Public Welfare Scenarios

Tencent launched the public welfare industry version of WorkBuddy, with 15 intelligent agents covering 42 public welfare scenarios.

【AiBase Summary:】

🧠 Public Buddy integrates 11 scenario capsules and 42 scenario capabilities, supported by 15 intelligent agents.

📊 Public Buddy can generate daily reports, weekly reports, monthly reports, analyze donor data, and handle invoices.

🤝 Tencent SSV provides points funding to encourage institutions to accumulate public welfare skills and open them to the industry.

8、Anthropic Releases Claude Fable5.1: Stronger Agent Coding Ability, Up to 45% Cost Reduction

Anthropic launched the new AI models Claude Fable5.1 and Mythos5.1, significantly improving the coding and scientific reasoning capabilities of agents and optimizing reasoning costs. Fable5.1 opens up API access to public cloud platforms, while Mythos5.1 is restricted to specific projects. The new models include AI text watermarks and have optimized safety filtering mechanisms, reducing false positives. Cost-effectiveness is the core of this upgrade, with Fable5.1 saving about 25% in costs by lowering cache read prices, and complex workflows saving up to 45%.

【AiBase Summary:】

🧠 Claude Fable5.1 performs strongly in agent tasks, scoring 55.8% on Terminal-Bench4.0 test.

🔬 Mythos5.1 scores 52.6% in scientific research tests, doubling the performance of the previous Fable version.

💰 Fable5.1 saves about 25% in costs by lowering cache read prices, and complex workflows save up to 45%.