# large language models
Latest news and articles about large language models
Total: 23 articles found
Alibaba Unveils 'Wukong' — A Production-Grade Agent Platform for Enterprises
Alibaba has launched Wukong, an enterprise-grade Agent platform designed to coordinate large language models with business systems to automate multi-step tasks. The move deepens Alibaba’s push to sell higher-margin, production-ready AI orchestration to enterprises but raises questions about reliability, governance and cross-border compliance.
China’s AI Reckoning: CCTV’s ‘Model Poisoning’ Exposé, Tesla’s Gigawatt Chip Push and a Sprint to Robot Production
A CCTV consumer‑rights investigation revealed deliberate “poisoning” of AI model inputs via fabricated product content, exposing weaknesses in model provenance and retrieval. Simultaneously, Tesla announced a rapid build‑out of an in‑house wafer fab for next‑generation AI chips, while Chinese robot makers showcased rapid mass‑production progress and retail partnerships at AWE 2026.
How Cheap “GEO” Services Are Teaching Chinese AI Models to Lie
China’s 315 consumer‑rights programme exposed how low‑cost “GEO” services buy visibility in AI recommendation pipelines by mass‑publishing fabricated content. The practice exploits retrieval behavior in deployed models, turning marketing budgets into a way to manufacture apparent evidence and influence consumer decisions, and has prompted regulatory scrutiny.
Baidu Cloud Unveils 'DuClaw' — A Zero‑Deployment Agent Service That Tethers Search to Large Models
Baidu Intelligent Cloud launched DuClaw, a zero‑deployment OpenClaw service that preloads Baidu search, Baike and academic search skills and supports multiple mainstream large models. The product aims to speed enterprise agent adoption by tightly coupling knowledge retrieval with LLMs, but it also raises questions about platform control and security oversight.
China’s Zhou Hongyi: Multi‑Agent AI, Not Chatbots, Will Unlock Real‑World Deployment
Zhou Hongyi, founder of 360, warned that large models today are essentially chatbots and argued that industrial deployment of AI requires converting these models into autonomous agents that can coordinate with one another. He urged a focus on multi‑agent systems, infrastructure and integration to unlock real‑world applications across industries.
Architect of Alibaba’s Qwen Steps Down Abruptly, Raising Questions for China’s Open‑Source AI Drive
Lin Junyang, a leading architect of Alibaba’s open‑source Qwen models, announced his sudden resignation on March 4, saying he needed to rest. His exit raises questions about continuity at a flagship Chinese AI programme even as Alibaba signals the team should continue with existing plans.
Alibaba’s Star AI Engineer Steps Down as Qwen Project Faces Leadership Gap
Lin Junyang, the technical head of Alibaba’s Tongyi Qianwen project and the company’s youngest P10 technical leader, announced his departure on March 4. Lin led Qwen’s rise as a prominent open-source LLM family, but Alibaba has not named a successor, leaving short-term uncertainty about leadership and the project’s future direction.
OpenAI Tweaks ChatGPT’s ‘Instant’ Brain with GPT‑5.3 — And Promises GPT‑5.4 Even Sooner
OpenAI has launched GPT‑5.3 Instant, an update designed to make ChatGPT’s most used mode more helpful by reducing unnecessary refusals and improving context comprehension. The company also signalled a faster timetable for GPT‑5.4, suggesting more substantial capabilities — and more intense scrutiny — lie ahead.
From Megvii's Ashes to a Geely-Backed Empire: Yin Qi's Bet on an AI+Car Closed Loop
Yin Qi has reconfigured his post‑Megvii career around a tightly integrated AI+automotive strategy, combining StepFun’s large models with Qianli/Geely’s hardware and distribution to create a commercial ‘‘closed loop.’’ Backed by state capital and big tech, StepFun has raised over RMB5 billion and aims for a near‑term listing, but its dependence on Geely creates strategic trade‑offs between fast deployment and independence.
Baidu’s High-Stakes AI Gamble: Massive Write-Downs, Shrinking Ads and an Uncertain Payoff
Baidu’s 2025 results reveal a painful transition from search advertising to AI-driven products: revenue fell to ¥129.1 billion and net profit plunged 76% after a ¥16.2 billion impairment on obsolete compute. While AI revenue grew strongly, it has not yet replaced lost ad income and the company faces intense competition and uncertainty over AI unit economics.
Nvidia Targets the ‘Inference’ Bottleneck with a New Generation of AI Chips
Nvidia is designing a new class of chips optimized for AI inference, prioritizing latency, throughput and energy efficiency for real‑time model serving. The move aims to lower the cost of running large models at scale and strengthens Nvidia’s position across the AI value chain while intensifying competitive and geopolitical pressures in the semiconductor industry.
China Overtakes US in AI Calls as Four Homegrown Models Dominate Global Top Five — Experts Point to Inference-Efficiency Strategies
NetEase reports that China’s aggregate AI API call volume has surpassed the United States for the first time, with four Chinese large models filling four of the global top five usage slots. Experts attribute the surge largely to engineering choices that reduce inference costs, enabling mass deployment across consumer and enterprise services.