Intelligence Archive
Historical Record / Total 404 Articles
May 2026
NetEase Youdao Open-Sources Confucius 4 Multimodal and TTS Models
NetEase Youdao has fully open-sourced the core dual engines of its Confucius 4 large model: a 27B-parameter multimodal model achieving SOTA on visual math problems, and a TTS model supporting cross-lingual emotion cloning across 14 languages. Both models are now available for free download and commercial use.
DeepSeek-V4-flashChina to Accelerate Embodied AI Training Infrastructure for Industrial and Consumer Deployment
China's National Development and Reform Commission announces plans to fast-track embodied intelligence training infrastructure, aiming to move humanoid robots beyond laboratory competitions and into factories, retail environments, and households.
KIMI - K2.5《人工智能应用伦理安全指引 1.0》正式发布，清华、阿里、华为等联合起草
全国网络安全标准化技术委员会在2026年中国网络文明大会上发布《人工智能应用伦理安全指引 1.0》，旨在引导AI应用以人为本、智能向善，推动行业规范发展。清华大学、阿里巴巴、华为等多家机构参与起草。
XIAOMI Mimo-v2-flashTencent Meeting Rolls Out AI Simultaneous Interpretation with Voice Cloning and Sub-3-Second Latency
Tencent Meeting has launched an AI-powered simultaneous interpretation feature that mimics user voices, delivers translations with less than three seconds of latency, and integrates seamlessly with real‑time transcription and captions.
DeepSeek-V4-ProJeff Bezos Dismisses AI Anxiety: Technology Will Elevate Human Value
Amazon founder Jeff Bezos urges software engineers not to fear AI, comparing the technology to a 'bulldozer' that enhances productivity rather than a replacement for human labor, despite widespread industry concerns regarding job displacement.
GLM-5Sberbank Targets Chinese AI Chips for GigaChat Compute Power
Russian banking giant Sberbank seeks to procure Chinese-manufactured AI chips to support its GigaChat model, facing potential supply constraints due to high domestic demand within China.
GLM-5月之暗面拆除VIE架构冲刺香港IPO，估值超200亿美元
据科技媒体Digitimes报道，中国AI初创公司月之暗面（Kimi）正拆除VIE与红筹结构，为赴港上市铺路。公司估值已超200亿美元，近期完成20亿美元融资。
XIAOMI Mimo-v2-flashOpenAI Introduces Dual Provenance Measures for AI-Generated Images: Invisible Watermarks and C2PA Metadata
OpenAI has announced a two-pronged approach to trace AI-generated images by embedding C2PA metadata and SynthID invisible watermarks, aiming to enhance authenticity and combat misuse.
DeepSeek-V4-flashAlibaba's Qwen3.5-LiveTranslate Achieves 2.8-Second Latency for Real-Time Speech Translation
Alibaba's Qwen3.5-LiveTranslate-Flash reduces end-to-end latency to 2.8 seconds, supports 60 input and 29 output languages, and introduces real-time voice cloning and dynamic hotword engines for cross-border live streaming, meetings, and more.
DeepSeek-V4-flashMarch 2026
Alibaba's Qianwen App Offers 5 Yuan Discount on Mobile Recharges Using Alipay 'AI Pay'
Alibaba's Tongyi Qianwen AI assistant has integrated Alipay's 'AI Pay' feature, enabling voice-activated mobile phone top-ups with a limited-time 5-yuan discount available through April 6.
KIMI - K2.5Google TurboQuant Algorithm Reshapes AI Efficiency, Sending Storage Chip Stocks Plummeting
Google researchers have introduced a new algorithm called TurboQuant, which drastically reduces AI memory usage by 1/6 and boosts inference speeds by 8x. The technology has caused a market crash in the storage chip sector, with major companies seeing significant stock drops.
GLM-4.7-FlashMoonshot AI Founder Yang Zhilin Cites Algorithm Scaling for Stable Pricing, Outlines AI Agent Adoption Roadmap
At the 2026 Zhongguancun Forum, Moonshot AI founder Yang Zhilin confirmed the company has not raised prices thanks to algorithmic cost reductions, while predicting phased AI agent adoption from early adopters to universal productivity penetration.
KIMI - K2.5