AI Morning News: Alibaba’s Qwen3.8 with 2.4 Trillion Parameters to Launch; Ant Digital Unveils Agentar 2.0, a Super Factory for Business Agents

Avatar 0

Alibaba’s Qwen3.8 Is Almost Here, and It’s a 2.4-Trillion-Parameter Beast

July 19th – Word on the street is that Alibaba’s latest large language model, “Qwen3.8,” is about to drop officially. This model boasts a whopping 2.4 trillion parameters, and the company is calling it one of the most powerful models out there right now, falling just short of Fable 5 in performance. It’s also built to play nice with all the major cutting-edge AI architectures. As of now, the Qwen3.8-Max preview version has already landed on Alibaba’s Token Plan, Qoder, and QoderWork platforms, so early birds can get a taste.

SenseTime Rolls Out the SenseNova U1 Pro, a Model Built for Real-World Tasks

July 18th – At the 2026 World Artificial Intelligence Conference (WAIC), SenseTime dropped its new model, the SenseNova U1 Pro. From what we’ve gathered, this model is designed as a native multi-modal intelligent agent foundation for long-horizon tasks. It’s all about tackling the “repeated card-drawing” headache that plagues many multi-modal products, pushing the industry from “one-off content generation” to a “system-level content delivery” phase where it can actually close out complex, long-winded tasks.

On the tech side, the U1 Pro’s core is built on the NEO-unify kernel architecture, which creates a unified representation for both language and vision within a native multi-modal framework. Right now, the preview version is open for invited testing, and the full version is expected to launch this August, complete with API services.

Ant Digital Launches Agentar 2.0: A Super Factory for Business Agents

July 19th – Also at the 2026 WAIC, Ant Digital officially unveiled “Agentar 2.0,” a super factory platform for building commercial intelligent agents. Agentar 2.0 comes with an initial batch of 200 pre-loaded job-specific digital expert templates and offers hundreds of Skills-level, ready-to-use agent tools that you can subscribe to.

Alibaba Cloud’s Bailian Platform Goes Live with HappyOyster 1.0, an Open World Model

July 19th – Alibaba Cloud’s Bailian platform announced the launch of HappyOyster 1.0, an open-world model product that lets you build and interact with worlds in real time. According to the team, HappyOyster 1.0 has two main modes: Adventure (for world exploration) and Directing (for real-time scene control). Just type in a sentence or upload a picture, and it’ll generate an interactive, explorable AI digital world. It already offers SDKs for Android, iOS, and Web, and is currently in grey testing on Alibaba Cloud Bailian.

Tencent Cloud Unveils ADP 4.0 International Edition

July 18th – At the 2026 WAIC, Tencent Cloud officially released the international version of its Tencent Cloud Agent Development Platform, ADP 4.0. This update also upgrades three core modules: the intelligent workbench, the Claw mode, and the Skill Square.

NetEase Youdao Shows Off OpenPods, an AI-Powered Headset at WAIC

July 18th – Over at the 2026 WAIC, NetEase Youdao gave the public its first look at the Youdao OpenPods, an AI-powered pair of earbuds. These things support real-time translation for over a dozen languages, and they can also transcribe recordings, automatically generate structured meeting minutes, and even create mind maps.

Word is, these earbuds have already gone into mass production and are expected to make their official debut in August. Youdao says they plan to take the OpenPods global.

Bouchou Quantum Debuts “Liangchou No.1,” an Atomic Quantum AI Base

From July 17th to 20th, the 2026 World Artificial Intelligence Conference and High-Level Meeting on Global AI Governance (WAIC 2026) was held in Shanghai. At the event, Shanghai Bouchou Quantum Technology Co., Ltd. launched its first atomic quantum artificial intelligence base (Q-Cub), the “Liangchou No.1.” They also introduced three configuration tiers: a Cloud-Linked entry-level version (with a shared QPU) for university research, a standard Appliance version (with a private QPU+GPU) for corporate R&D centers and industry benchmark projects, and a flagship Infrastructure version (with a full trio of computing clusters) for state-owned enterprises, large institutions, and joint labs. In addition, Bouchou Quantum, together with the Shanghai Qi Zhi Research Institute and Fudan University, announced the establishment of the “Neutral Atom Quantum Intelligence Laboratory.”

Advertisement

Leave a Reply

Your email address will not be published. Required fields are marked *

Log In / Sign Up

Enter your email to receive a secure code. No password needed.