- Apple Unveils Full AI Overhaul at WWDC 2026, Cook Bids Farewell as CEO
At the Worldwide Developers Conference (WWDC) 2026 held in Cupertino on June 8, Apple launched its long-awaited next-generation Apple Intelligence ecosystem and a fully rebuilt Siri AI, alongside updated iterations of iOS 27, iPadOS 27, and macOS 27. The new Siri has transformed from a fixed-command voice assistant into an all-capable AI conversational helper that supports cross-platform continuous tasks across Apple’s core operating systems and understands personal context to retrieve information from emails, photos and apps. CEO Tim Cook hosted the conference for the last time before stepping down; he will transition to executive chairman in September, with hardware engineering head John Ternus set to take over the CEO role. Market reaction was muted: Apple’s stock closed 1.9% lower on June 8, erasing $85 billion in market value, with all new AI functionalities currently unavailable in the Chinese market pending regulatory approval.
- Chinese AI Large Models Surpass US in Weekly Call Volume, Claim 4 of Top 5 Spots Globally
Latest data from OpenRouter shows that between June 1 and 7, total weekly token calls for global AI large models reached 36.1 trillion, up 13.5% week-on-week and marking the 7th consecutive weekly increase. Chinese AI large models recorded 14.19 trillion token calls, a 27.49% jump from the previous week, and have outperformed US models for six straight weeks to hold the top global position. In the top 5 most-called models worldwide last week, four were Chinese-developed, with DeepSeek-V4-Flash retaining the number one spot with 3.69 trillion weekly tokens.
- OpenAI Launches GPT-5 with Cutting-Edge Reasoning and Multimodal Capabilities
OpenAI has officially unveiled GPT-5, its next-generation flagship large model, which delivers dramatically improved logical reasoning, expanded context windows, and enhanced multimodal understanding. The new model has set new state-of-the-art records on math, coding, and scientific research benchmarks. Early access is currently rolling out to ChatGPT Plus subscribers and API customers, marking OpenAI’s most significant model upgrade since 2024.
- Amap Releases World’s First Native 3D Global City World Model ABot-Earth 0.5
Chinese tech giant Alibaba’s mapping subsidiary Amap launched ABot-Earth 0.5, the world’s first natively trained 3D usable city world model, on June 8. The model already covers 190+ countries and regions, building the largest global coverage 3D map service in the world. Unlike traditional 3D modeling that distorts 3D structures from 2D images, ABot-Earth 0.5 trains directly on native 3D data; it can generate kilometer-level editable 3D city scenes in 10 minutes on a consumer-grade GPU from just one satellite image or text description. Amap says the new approach cuts 3D mapping costs to just 1% of traditional methods, while boosting efficiency by around 1,000 times.
- Tencent Opens All Hunyuan AI Agents to Small and Medium Enterprises for Free
During Tencent’s recent AI industry summit, Tencent announced that its full suite of Hunyuan-based industry intelligent agents will be available for free to small and medium-sized enterprises (SMEs) for three months, accelerating industrial AI landing across domestic sectors. The move comes as Tencent reshuffles its AI business structure and hires former OpenAI researcher Shunyu Yao as its chief AI scientist, ramping up its push into the AI agent era after years of accumulation in scenario, connection and engineering capabilities.
- Ideogram 4.0 Open-Sourced, Claims Top-Tier Text-to-Image Capabilities with 9.3 Billion Parameters
AI image startup Ideogram has open-sourced Ideogram 4.0, a 9.3-billion-parameter text-to-image model that ranks fourth globally on the AI DesignArena benchmark. The release allows developers to download full model weights, fine-tune the model with custom data, and run it on local on-premises hardware, marking a key shift in the image generation market from closed cloud services to open-weight, deployable solutions for enterprise users. The model is now available across all Ideogram subscription plans and via public API.
- StepFun Prepares Hong Kong’s Largest AI IPO in Recent Years
Chinese AI startup StepFun (Jieyue Xingchen), founded by former Microsoft corporate vice president Daxin Jiang in 2023, is set to submit its IPO application to the Hong Kong Stock Exchange as early as this week, with a top valuation of up to $120 billion. If successful, the offering will become the largest AI IPO in Hong Kong in recent years, leading a new wave of public listings for Chinese large model developers.
- OpenAI Upgrades ChatGPT Memory System, Cuts Computing Costs to One-Fifth
OpenAI rolled out a major upgrade to ChatGPT’s memory system, targeting two core pain points of outdated stored information and frequent memory errors. The new architecture reduces the computing power required for personalized memory storage and retrieval to just one-fifth of the previous system, delivering more stable long-term personalized interaction for users while cutting operational costs for the company.
- xAI Releases Grok Imagine Video 1.5, Enabling One-Click Image-to-Video Conversion
Elon Musk’s AI company xAI launched Grok Imagine Video 1.5, which can generate a full smooth video from a single input image in just a few seconds. The new product directly competes with Google’s recently released Veo video generation model, intensifying the race in the fast-growing text/image-to-video AI track.
- China’s National Data Administration Launches Six Major Initiatives for High-Quality Industry Datasets
On June 8, China’s National Data Administration released an action plan to promote the construction of high-quality industry datasets, rolling out six targeted initiatives covering the full supply chain of high-quality datasets: capacity expansion, annotation upgrading, quality improvement, application empowerment, management services, and value release. The plan aims to build a batch of validated, industry-focused high-quality datasets by 2028, forming a closed industrial loop of “scenario-driven data, data empowering models, models enabling applications, and applications creating value”, marking a key step in building a standardized data foundation for China’s AI industry development.