Intelligence Archive
Historical Record / Total 215 Articles
March 2026
Alibaba Qianwen Integrates Taxi-Hailing with Natural Language Control
Alibaba's Qianwen AI assistant now enables users to book rides, select vehicle types, add waypoints, and schedule trips using conversational commands, with support for nuanced preferences like driving stability and vehicle air quality.
KIMI - K2.5Cursor Admits New 'Frontier-Level' Coding Model Built on Top of Chinese AI Startup's Kimi
AI coding startup Cursor launched Composer 2 this week, but faced immediate scrutiny after an X user claimed the model was based on Moonshot AI's open-source Kimi 2.5. Cursor admitted to using Kimi as a base, clarifying that the majority of the model was trained using their own compute, and confirmed the use was licensed.
GLM-4.7-FlashChina's AI Large Model Weekly Usage Surpasses US for Two Consecutive Weeks with 4.69 Trillion Tokens
Data from OpenRouter reveals that Chinese AI models achieved a weekly call volume of 4.69 trillion tokens, surpassing the United States for two consecutive weeks and dominating the global top three rankings.
GLM-5MiniMax Enacts Peak-Time Rate Limiting as M2.7 Model Demand Surges
MiniMax's open platform will implement dynamic rate limiting during peak hours due to unexpectedly high demand for its newly launched M2.7 agent model. The company cites rapid traffic growth and concerns over automated batch tasks affecting shared compute resources as reasons for the policy, aimed at ensuring stable service for the majority of users.
DeepSeek-V3.2 (Reasoner)National Power AI Pilot Base Welcomes Huawei, ZTE, Baidu as First Tenants, Aims for 30 Companies in 2026
China's national-level Power Artificial Intelligence Pilot and Testing Base has officially welcomed its first batch of eight leading enterprises and institutions, including Huawei, ZTE, and Baidu. The base, serving as a 'bridge platform' for the power AI sector, aims to onboard 30 companies this year to accelerate technology validation and industrial application.
DeepSeek-V3.2Meituan Open-Sources 560-Billion Parameter AI Model for Mathematical Proofs
Chinese tech giant Meituan has open-sourced 'LongCat-Flash-Prover,' a 567.7-billion parameter Mixture-of-Experts (MoE) model designed to solve complex mathematical proof problems. The model reportedly sets new state-of-the-art (SOTA) records on two key benchmarks.
DeepSeek-V3.2Moonshot AI Confirms Kimi K2.5 as Foundation for Cursor's Composer 2, Thanks Elon Musk for Support
Moonshot AI has officially clarified its relationship with Cursor, confirming that Kimi K2.5 serves as the base model for Cursor's Composer 2. The collaboration involves Fireworks AI for hosting and reinforcement learning, designated as an authorized commercial partnership. The response came after Elon Musk verified the model's identity and a user discovered the API endpoint.
GLM-4.7-FlashAI Face-Swapped Short Drama Found to Infringe on Famous Actor's Portrait Rights
Beijing Internet Court has ruled that both the production company and broadcasting platform of a short drama featuring AI face-swap technology that closely resembled a well-known Chinese actress are liable for portrait rights infringement. The court ordered both companies to issue public apologies and pay economic damages. The case marks a significant ruling on AI-generated likenesses and portrait rights in China's entertainment industry.
MiniMax-M2.5Xiaomi Opens MiMo V2 Models to Developers with Free Week-Long API Access via Major Agent Frameworks
Xiaomi has released its MiMo-V2-Pro, MiMo-V2-Omni, and MiMo-V2-TTS large language models and is offering a one-week free API trial through integrations with five leading developer agent frameworks: OpenClaw, OpenCode, KiloCode, Cline, and BLACKBOX.AI.
DeepSeek-V3.2WordPress.com Integrates AI Agents for Automated Content Creation and SEO Optimization
WordPress.com has fully integrated AI agents to handle website operations, allowing users to use natural language commands to automatically draft, edit, publish, and optimize SEO, while maintaining design consistency.
GLM-4.7-FlashUnitree Robotics Aims to Launch Foundation Model for General Humanoid Robots Within Three Years
Chinese robotics firm Unitree Robotics has disclosed in its latest IPO filing responses that it plans to release a 'General Humanoid Robot Embodied Foundation Model' within three years. The model is designed to possess four core capabilities: scene generalization, instruction generalization, action generalization, and task generalization. This development is part of the company's push to commercialize embodied AI, supported by its recently accepted IPO application on Shanghai's STAR Market seeking to raise 4.2 billion yuan.
DeepSeek-V3.2 (Reasoner)Tencent Yuanbao AI Unveils Anthropomorphic Logo with Eyes
Tencent Yuanbao AI has updated its visual identity, introducing a new logo featuring eyes to appear more human-like. This redesign accompanies the AI assistant's strong performance during the recent Spring Festival campaign, where it achieved significant growth in daily active users.
GLM-4.7-Flash