Intelligence Archive
Historical Record / Total 372 Articles
May 2026
Zhiyuan's 2B-Parameter World Model GE 2.0 Tops WorldArena, Outshining Nvidia and Microsoft
Zhiyuan AGIBOT's self‑developed world model GE 2.0, with only 2 billion parameters, has claimed first place in the WorldArena Track1 benchmark for embodied AI perception and action. The lightweight model outperformed heavy‑duty competitors from Nvidia, Microsoft, and a Tsinghua–Stanford team, demonstrating near‑real‑time long‑horizon generation and strong real‑world correlation without specialized benchmark tuning.
DeepSeek-V4-ProXiaomi Open Sources ControlFoley Model for Controllable Video Sound Generation
Xiaomi's Large Model Application Team has released ControlFoley, an open-source framework designed to solve controllability challenges in video sound effect generation, achieving state-of-the-art results across multiple benchmarks.
GLM-5Anthropic's Claude Opus 4.8 Accused of Distilling Chinese Models in 'Identity Crisis' Controversy
Shortly after the release of Claude Opus 4.8 and a massive $65 billion funding round, users discovered the model identifying itself as Chinese AI models Qwen and DeepSeek via API tests, sparking allegations of hypocrisy regarding data distillation.
GLM-5StepFun Open-Sources Step 3.7 Flash Model with 400 Tokens/s Generation Speed
StepFun has released Step 3.7 Flash, a sparse MoE model optimized for agent workflows, coding, search, and multimodal tasks. Delivering up to 400 tokens per second, it supports native multimodal understanding, web/visual search, and reliable tool calling across major agent frameworks.
DeepSeek-V4-flashiFlytek Launches AI Glasses with 122-Language Translation, Priced at $590
iFlytek unveiled its AI glasses at BEYOND Expo 2026, featuring real-time translation across 122 languages, a lightweight 40g design, and built-in AI assistant. Pre-orders start June 15 at ¥4,299.
DeepSeek-V4-flashChina Unveils 2026 Digital Literacy Roadmap with Focus on National AI Competency
Four major Chinese ministries jointly issued the "2026 Work Points for Improving Digital Literacy and Skills," outlining 15 tasks across six pillars to boost AI education, accelerate talent cultivation, and ensure inclusive digital development.
KIMI - K2.5Xiaomi Slashes MiMo API Prices by Up to 99%, Citing Structural Cost Advantages
Xiaomi has announced a permanent reduction of up to 99% for its MiMo-V2.5 series API prices. MiMo lead Luo Foli attributes the aggressive pricing to architectural optimizations that keep inference costs far below industry averages, allowing the division to break even while stimulating broader AI infrastructure development.
GLM-5Stardust Intelligence Unveils T1 Rope-Driven Robot: Capable of Cooking and Laundry for 89,900 RMB
Stardust Intelligence has officially released the T1 rope-driven robot, a versatile machine designed for both household and commercial tasks. Priced from 89,900 RMB, the T1 stands 1.55 meters tall and features 23 degrees of freedom, capable of cooking, folding laundry, and even performing chemical experiments.
GLM-5Tencent Hunyuan Unveils Hy-Memory: A 'Second Brain' for Long-Term AI Agents
Tencent Hunyuan has officially launched Hy-Memory, a specialized memory plugin for the Openclaw agent. Designed to solve the 'three-week trajectory' of user frustration, the system utilizes a 6-layer framework and a dual-system cognitive architecture to ensure agents 'remember correctly' and evolve with users over time.
GLM-4.7-FlashByteDance Reportedly Developing Custom CPUs to Fuel AI Infrastructure Expansion
Facing rising costs and supply shortages, ByteDance is developing its own CPUs based on Arm and RISC-V architectures to support its growing AI infrastructure and推理 (inference) workloads.
GLM-5ByteDance Eyes Up to $70 Billion in AI Infrastructure Spending
ByteDance is reportedly preparing to spend up to $70 billion this year on data centers and AI infrastructure, a massive increase from previous years, aiming to solidify its global lead in AI technology.
GLM-4.7-FlashHuawei's Zheng Jun: AI Model Gap Narrowed to 2.7%, Chinese Usage Surpasses US
At the 2026 Phoenix Bay Area Financial Forum, Huawei's Zheng Jun stated that China's AI models are only 2.7% behind the US in performance. He noted that Chinese model usage has surpassed US usage since February 2026, attributed to open-source advancements and cost efficiency. The article also covers policy support, data growth, and infrastructure advantages.
GLM-4.7-Flash