AI Daily: Qwen 3.5-Max Model Released; Doubao's AI E-commerce Internal Testing; Tencent QClaw Officially Opens Public Beta

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 innovative AI product applications.

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

1. China's Model Rises to the Top! Alibaba Tongyi Qianwen 3.5-Max Makes a Stunning Debut: Global Ranking Jumps to Fifth

Chinese large models once again set new records on the international top stage. Alibaba Tongyi Qianwen 3.5-Max-Preview achieved a comprehensive score of 1464 points, ranking fifth on the LM Arena global large model company ranking list, demonstrating the strength of domestic large models in global competition.

AiBase Highlights:

🧠 Mathematical ability ranks fifth globally, showcasing strong logical reasoning capabilities.

⚙️ Comprehensive performance ranks sixth globally, performing excellently without style control.

📄 Expert-level processing capability ranks among the top ten globally, with strong text processing capabilities.

2. Ambition Behind 220 Million Monthly Active Users: Doubao Tests AI E-commerce, "One Sentence Order" E-commerce Loop

Doubao's internal testing of AI e-commerce features demonstrates its ambition in the fields of AI technology and e-commerce, while also revealing the intense competition in the AI e-commerce sector.

AiBase Highlights:

🛒 Doubao tests AI e-commerce features, achieving a "one-sentence shopping" experience.

📈 Doubao has 226 million monthly active users, leading the domestic AI-native application market.

🔄 The competition in AI e-commerce is intensifying, with top platforms having already achieved large-scale implementation.

3. Tencent QClaw Officially Opens Public Beta: No Invitation Code Needed, AI "Crayfish" for Remote PC Control via WeChat

Tencent QClaw officially opens public beta, as an AI assistant, its core function is to remotely control PCs through WeChat, providing efficient office and development support.

AiBase Highlights:

🦞 QClaw is an AI assistant launched by Tencent, supporting remote PC control via WeChat, and can be used without an invitation code.

💻 The product is built based on the open-source OpenClaw framework, emphasizing "WeChat direct connection + zero-barrier deployment," supporting Mac and Windows systems.

🌐 It provides a rich Skills ecosystem, including file processing, email generation, GitHub auto-submission, etc., covering office and development scenarios.

Details link: https://qclaw.qq.com/

4. Google Secretly Testing macOS Version of Gemini App, May Launch "Desktop Intelligence" Screen Sharing Feature

Google is actively expanding its AI ecosystem, launching an early Beta test of the native Gemini app for macOS. The app will introduce the "desktop intelligence" feature, allowing Gemini to deeply understand user desktop applications and screen content, thus enabling direct interaction with AI. Additionally, the app supports creating multimedia content and has context understanding and web search capabilities.

AiBase Highlights:

🍎 Google has started the early Beta test of the native macOS version of Gemini to compete with rivals.

🖥️ Introduces the "desktop intelligence" feature, enhancing Gemini's understanding and interaction with the user's desktop environment.

📷 Supports creating multimedia content, along with context understanding and web search capabilities.

5. Canva Plans to Go Public Next Year, Fully Transforms into an AI-Driven Design Ecosystem

Canva plans to start its IPO process next year, marking a key phase in its AI transformation journey. The company is committed to a comprehensive technological innovation of its design platform, shifting its strategic focus from traditional creative tools to an AI-driven content creation ecosystem. To enhance market competitiveness, Canva has recently launched a series of generative AI features, aiming to lower the professional design threshold for individual users and small businesses, using algorithms to achieve automated formatting, image generation, and creative collaboration.

AiBase Highlights:

🧠 Canva plans to initiate its IPO process, marking a key stage in its AI transformation.

🤖 Canva is undergoing comprehensive technological innovation, shifting from traditional design tools to an AI-driven content creation ecosystem.

💡 Canva lowers the design barrier through generative AI features, enhancing market competitiveness.

6. ElevenLabs Launches AI Music Trading Market, Pioneering the "Creator Revenue Sharing" Business Model

ElevenLabs launches an AI music trading market, marking its expansion of voice cloning monetization into the full audio field, and promoting the assetization of AI content through a creator revenue-sharing business model. However, the platform still faces copyright legal issues, requiring users to bear risks themselves.

AiBase Highlights:

🎧 The AI music trading market is officially launched, allowing users to create, publish, and sell AI-generated songs.

⚖️ The music licensing system is divided into three levels: social media, paid marketing, and offline use, covering diverse needs.

⚠️ The platform explicitly states that it does not guarantee the exclusivity of generated content and does not provide legal protection guarantees.

7. Cursor Releases Composer 2 Programming Model: Performance Close to GPT-5, Price Just a Fraction of It

Cursor released its self-developed programming large model, Composer2, which achieved a score of 61.3 on the CursorBench evaluation, surpassing Claude Opus4.6 and closely following GPT-5.4Thinking. Composer2 uses a highly restrained training strategy, focusing on code processing, and challenges OpenAI and Anthropic with a highly competitive price (only $0.50 per million Tokens). By releasing Composer2, Cursor aims to free itself from reliance on competitors' APIs, building a complete moat from editor to model base, ensuring a competitive advantage in the second half of the AI programming race.

AiBase Highlights:

🧠 Composer2 performance leaps forward, surpassing Claude Opus4.6 and closely following GPT-5.4Thinking.

💰 Composer2 is a cost-cutting model, reducing costs by up to 90%.

🔄 Cursor moves away from the "shell application" label, establishing a complete moat.

8. European AI Dominator Unveils "All-Rounder"! Mistral Small 4 Released: Reasoning, Multimodal, Programming, I Want It All

Mistral AI released Mistral Small4, a versatile large model with strong reasoning, multimodal understanding, and programming capabilities. It uses an advanced MoE architecture with 119B parameters and only 6B activated parameters, offering high operational efficiency. In terms of performance, Mistral Small4 has made a qualitative leap compared to its predecessor, with end-to-end completion time reduced by 40% under delay optimization mode.

AiBase Highlights:

🧠 Mistral Small4 is a versatile large model that can handle reasoning, multimodal understanding, and programming.

⚙️ It adopts an advanced MoE architecture, with 119B parameters and only 6B activated parameters, offering high operational efficiency.

🚀 In terms of performance, Mistral Small4 has made a qualitative leap compared to its predecessor, with end-to-end completion time reduced by 40% under delay optimization mode.