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. Claude Opus 4.8 Beats GPT-5.5! Launches with a Two-Thirds Price Drop
Anthropic's Claude Opus 4.8 has made significant improvements in agent programming, complex logic reasoning, and multi-domain knowledge work, while also optimizing user experience and computational costs.
AiBase Summary:
🧠 Enhanced agent programming and complex logic reasoning capabilities
📈 Outperforms GPT-5.5 and Gemini3.1Pro in multiple benchmark tests
⚡ Improves performance and reduces computational costs
2. Xiaohongshu PC Version Launches AI Search Assistant "Dian Dian": Supports Multi-turn Conversations and Collection Import Based on Massive Real Notes
Xiaohongshu launched the AI search assistant "Dian Dian" on its PC version, providing active conversational interaction based on a large volume of real notes, supporting multi-turn conversations and collection import, improving answer accuracy and personalized experience. This feature further complements multi-device scenarios and breaks through traditional search engine barriers through the differentiated approach of "generative AI + real community ecosystem," deepening its AI search strategy and reshaping content community traffic entry points and information distribution efficiency.
AiBase Summary:
🤖 Xiaohongshu launches AI search assistant "Dian Dian" on PC, supporting multi-turn conversations and collection import
🌐 Breaks through traditional search engine barriers with the differentiated approach of "generative AI + real community ecosystem"
📈 Deepens AI search strategy, reshapes content community traffic entry points and information distribution efficiency
3. Twice as Fast! Jieque Star Releases Open-Source Step 3.7 Flash Large Model
Jieque Star released and open-sourced the new Flash model Step 3.7 Flash, designed specifically for the production deployment of agents, with deep optimization for code writing, online search, and multimodal workflows.
AiBase Summary:
🧠 Uses an advanced sparse mixture-of-experts (MoE) architecture, with a total parameter count of 196B.
🌐 Possesses strong native multimodal understanding capabilities, directly identifying and parsing complex visual information such as UI interfaces, charts, and documents.
🔧 Demonstrates high stability in long-chain, multi-turn agent workflows.
4. AI Spreads to E-commerce Governance! Pinduoduo Launches Over 40 Measures in the First Half of the Year, with High-tech Cheating as a Key Target
Pinduoduo has recently increased its efforts to govern AI-related illegal products and services, especially in fields related to exams, such as AI question prediction tools and covertly disguised services, which are being targeted for removal. At the same time, the platform has also strengthened the governance of AI-generated content identification and introduced multiple comprehensive compliance measures to enhance the safety and health of the consumer shopping environment.
AiBase Summary:
🧠 Pinduoduo is governing AI-related illegal products and services, targeting AI question prediction and data poisoning.
🔐 The platform has enhanced compliance requirements, implementing special governance for AI-generated content identification to comply with the "Artificial Intelligence Generated Synthetic Content Identification Method."
🛒 The platform has introduced over 40 comprehensive governance measures in the first half of the year, covering core areas such as product access, advertising governance, and intellectual property protection.
5. Grok Build 0.2.7 Major Update: /usage, /login Launched, Multi-Sub-Agent Shared Terminal + Image Understanding Boosted
The Grok Build 0.2.7 update brings several core improvements, including the addition of /usage and /login commands, optimizing user management and login processes, while introducing a multi-sub-agent shared terminal function, significantly enhancing image understanding capabilities and cross-platform compatibility, offering developers a more efficient coding experience.
AiBase Summary:
/usage command launched, improving user management convenience.
/login support optimized, enhancing account switching experience.
Multi-sub-agent shared terminal function, increasing parallel task processing capability.
Details link: https://x.ai/build/changelog
6. Xiaomi Announces MiMo-V2-Pro/Omni Will Be Discontinued in June 2026, Fully Migrated to V2.5 Series
Xiaomi announced that it will discontinue the old version of MiMo models on June 30, 2026, and replace them with the MiMo V2.5 series. This move aims to improve reasoning capabilities and cost-effectiveness, while encouraging developers to migrate to the new version.
AiBase Summary:
🚀 Xiaomi will discontinue MiMo-V2-Pro/Omni models, replacing them with the MiMo V2.5 series.
💡 The new version model offers stronger reasoning capabilities and higher cost-effectiveness.
🔄 Xiaomi improves developer call efficiency and platform stability by unifying model versions.
7. Alibaba Cloud BaiLian Fully CLI-Enabled and Open-Sourced: One Command Connects Full Stack AI Agent Capabilities
Alibaba Cloud BaiLian is fully CLI-enabled and open-sourced, connecting full-stack AI Agent capabilities with one command, achieving a strategic shift from single-model calls to complex multi-capability collaboration, while lowering the threshold for developers to build complex task flows.
AiBase Summary:
🧠 BaiLian CLI encapsulates core capabilities like mainstream models and application workflows into a command-line interface
🌐 Achieves a strategic shift from single-model calls to complex multi-capability collaboration
🔧 Open-source initiative lowers the threshold for developers to build complex task flows
8. Memory Density Increases by 45%, Token Usage Drops by 35%!! Tencent Huan Yuan Launches Strongest Memory Plugin Hy-Memory
The article introduces the Hy-Memory intelligent memory plugin launched by Tencent Huan Yuan, which solves the problems of fragmented memory and chronological confusion in long-term collaborative AI agents through three core technologies. Hy-Memory performs exceptionally well on multiple test sets, significantly improving memory density and processing efficiency, and provides various configurations to suit different needs.
AiBase Summary:
🧠 Six-layer memory framework accurately positions memory, improving memory efficiency
⚙️ System1/System2 dual systems balance speed and depth
🔄 Evolutionary chain mechanism preserves causal traces, avoiding historical errors
