AI Daily: Qwen3.8-Flash Launched for Office; Zhipu Opensources Multimodal Large Model GLM-5.3-Flash; JD Superbrain 3.0 Large Model Makes Debut

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 technical trends and innovative AI product applications.

Fresh AI products Click to learn more: https://app.aibase.com/zh

1. Qwen3.8-Flash Launches on Tongyi Office, Marking the Beginning of the "Abundant and Efficient" Era for Agents

The Tongyi Office platform has officially launched the Qwen3.8-Flash large model, introducing a dual supply system of standard mode and advanced mode. The model has made breakthroughs in performance, cost, and efficiency, offering users a more efficient office experience.

AiBase Highlights:

🧠 Qwen3.8-Flash uses a new architecture, surpassing Claude Opus 4.6 in performance.

💡 Standard mode covers 95% of daily office needs, while advanced mode handles complex tasks.

🚀 Through collaboration between agents and large models, it achieves high performance, low cost, and low latency.

2. Price Reduced to 1/40th! Zhipu Opensources Native Multimodal Large Model GLM-5.3-Flash

Zhipu has open-sourced GLM-5.3-Flash, promoting the popularization of AI technology with high performance and extremely low pricing. It performs well on multiple platforms and has strong visual integration capabilities, providing developers with an easy-to-use experience.

AiBase Highlights:

🧠 GLM-5.3-Flash is the first native multimodal model, with leading performance and extremely low pricing.

💰 The price is only one-tenth of the same series, greatly lowering the usage threshold.

🚀 Architectural innovation using a hybrid sparse and linear attention mechanism improves inference efficiency.

More details: https://chat.z.ai

3. Compressed from 3.3GB to 440MB! How Does Tencent's Huan Yuan Ultra-Low Bit Translation Model Achieve Almost Zero Loss?

Tencent Huan Yuan team successfully compressed a 1.8B translation model to hundreds of megabytes, achieving almost lossless translation quality and successfully applied it to real-time translation of Bilibili live chat messages, improving the operational efficiency of edge devices and user privacy protection.

AiBase Highlights:

🧠 The Tencent Huan Yuan team successfully compressed a 1.8B translation model to hundreds of megabytes, achieving almost lossless translation quality.

⚙️ Using a 2-bit and 1.25-bit ultra-low bit quantization scheme, it improves model operation efficiency on edge devices.

🌐 In practical application, the model has been successfully used in real-time translation of Bilibili live chat messages, ensuring user privacy security.

4. Copying AI Software to Start a Business, Beijing's First AIGC Commercial Secret Case Fined 100,000 Yuan

The Chaoyang District Market Supervision Bureau of Beijing made a decision on the first administrative penalty case in the AIGC industry involving commercial secrets, issuing a fine of 100,000 yuan for an employee who privately copied and used the original unit's AI commercial software, marking a substantive step in Beijing's commercial secret protection in the new digital economy model.

AiBase Highlights:

💻 An employee secretly copied and cracked the original unit's AI commercial software and used it to start a new company for profit after leaving.

🔍 Law enforcement officers reconstructed the entire process of copying, cracking, and cross-company reuse through electronic data evidence collection, forming a complete chain of evidence.

⚖️ The involved party was fined 100,000 yuan and guided to improve confidentiality measures, marking the first confirmation that AIGC commercial software can be protected as commercial secrets.

5. JD's "Super Brain" 3.0 Large Model Makes Its Debut, Creating an AI Commander for the Entire Logistics Process

JD's "Super Brain" 3.0 large model has achieved industrial-level application integration of full-process AI command in the logistics industry, demonstrating powerful computational power and intelligent collaborative optimization across the entire supply chain, providing a practical and replicable industry model for the development of smart logistics.

AiBase Highlights:

🧠 JD's "Super Brain" 3.0 large model achieves industrial-level application integration of full-process AI command in the logistics industry.

🚀 "Super Brain" 3.0 demonstrates powerful computational power in core technologies and operational efficiency, capable of calculating the optimal path in seconds.

📦 "Super Brain" 3.0 connects the entire supply chain, covering key stages such as warehouse entry, transfer sorting, trunk transportation, and last-mile delivery.

6. FF Unveils Two New Robots: Next Futurist and Next Aegis, with Robot Factory Expected to Start Operation This Year

Faraday Future (FF) released two new robots, Next Futurist and Next Aegis, in Los Angeles, and announced the construction and product launch plans of its robot factory, while also showcasing its long-term strategic planning in the EAI field and AI technology.

AiBase Highlights:

🤖 FF unveiled the Next Futurist and Next Aegis robots, marking its strategic expansion from car manufacturing to embodied intelligence and EAI fields.

🏭 The robot factory will start operation this year, with the first product expected to be launched in February 2027.

💰 FF plans to invest no less than $5 billion over the next decade in EAI Robotics and AI, driving exploration of multimodal business models.

7. Google Upgrades Gemini Live Voice Function, Shifting AI Application Experience to "Complete a Task in One Sentence"

The article discusses Google's latest update to the Gemini Live voice function, aiming to allow users to use chat, Spark, and daily briefings more conveniently through natural voice commands. However, this design also highlights challenges in current AI application interactions, such as increased user understanding costs due to functional independence.

AiBase Highlights:

✨ Google launched the Gemini Live voice function, enhancing user convenience.

⚠️ Functional independence may increase user understanding and selection costs.

💡 The future direction of AI assistants is to reduce interaction complexity and achieve a unified entry point.

8. Claude Desktop App Cowork Now Has Built-in Browser, Capable of Completing Web Tasks Independently

Claude's desktop application now includes a built-in browser, allowing users to assign web tasks directly to it without installing extensions or additional settings. The built-in browser is completely isolated from regular browsers, ensuring security and supporting importing login information for specific websites from Chrome, Edge, and Firefox. This feature is rolled out to Pro, Max, and Team users, and Enterprise administrators can also enable it.

AiBase Highlights:

🟢 The built-in browser allows Claude to independently complete web tasks, improving work efficiency.

🟡 The built-in browser is completely isolated from regular browsers, ensuring user privacy and security.

🔴 Offers two browser options, suitable for different scenarios of web task requirements.