Flash

The news, the context, and what changes.

RSS
Flash

Meituan Open-Sources Digital Human Video Model LongCat-Video-Avatar 1.5

Meituan's LongCat team releases LongCat-Video-Avatar 1.5, significantly improving lip sync and video stability by upgrading the audio encoder, optimizing the data system, and introducing GRPO alignment. The model uses DMD distillation for 8-step generation, boosting efficiency by approximately 15 times, and advancing digital human technology from open-source SOTA to commercial-grade applications.

88Z AI editorial1 min read
Flash

NetEase Youdao Open-Sources Multimodal Large Model Confucius4

NetEase Youdao's AI team has open-sourced Confucius4, a multimodal reasoning model based on the Qwen3.5-27B architecture. Designed for advanced mathematical reasoning, it supports mixed text-image inputs and is compatible with the Qwen ecosystem.

88Z AI editorial1 min read
Flash

Alibaba's Production-Grade AI Digital Employee Platform QoderWake Launches Global Public Beta

Alibaba has launched QoderWake, a production-grade AI digital employee platform, initiating a global public beta with an initial Mac release. The platform allows for rapid configuration in five minutes and includes six built-in roles such as backend and frontend development. Featuring long-term memory and evolutionary capabilities, digital employees can accept tasks via IM or Webhooks and automatically generate delivery summaries. Security is ensured by a four-dimensional permission center, whil

88Z AI editorial1 min read
Flash

Zhipu Launches Next-Gen Large Model Inference Network Architecture ZCube

Zhipu, Yuxun Network, and Tsinghua University launched ZCube, an architecture that eliminates the Spine layer to address PD separation inference congestion. Tests show it significantly reduces hardware costs while boosting GPU throughput and response speed, providing an efficient foundation for ultra-large-scale inference clusters.

88Z AI editorial1 min read
Flash

Ali D20 Global Design Dean Summit Tickets On Sale, Duiyou Joins Again

Ticket sales have opened for the 2026 Ali D20 Global Design Dean Summit, with Duiyou participating once again. Centered on the theme of Design Rebirth, the main forum addresses large models, Agents, AIGC, and spatial intelligence. Sub-forums cover cutting-edge areas such as Alibaba Cloud Vibe Designing, Taotian AI Design Agent, and Qwen AI hardware UX, while introducing AI design workshops and exhibitions for the first time.

88Z AI editorial1 min read
Flash

Volcengine Agent Plan and Coding Plan Add DeepSeek V4

Volcengine Ark's Coding Plan and Agent Plan now support the DeepSeek V4 model series. Users can access both DeepSeek V4 Pro and V4 Flash within these packages. The Coding Plan is tailored for developer coding scenarios, while the Agent Plan integrates multimodal models like DeepSeek V4 and Doubao Seed, along with Harness capabilities such as web search. Subscription-based pricing significantly reduces invocation costs.

88Z AI editorial1 min read
Flash

Tencent Marvis Officially Launches, No Invitation Code Required

Tencent's OS-level AI assistant Marvis has officially launched, offering multi-platform downloads without an invitation code. Developed by the Yingyongbao team, it features always-on AI agents capable of reading local files, controlling PC software and mobile apps, and providing system-level file MCP services.

88Z AI editorial1 min read
Flash

Google I/O Developer Conference Unveils AI Suite

At I/O 2026, Google launched Gemini 3.5 Flash and Omni Flash models, upgraded search, introduced AI glasses and development platforms, marking the entry into an agent-driven Gemini era.

88Z AI editorial1 min read
Flash

Alibaba Qwen Launches New Flagship Model Qwen3.7-Max

Alibaba Tongyi Qwen has released the flagship model Qwen3.7-Max, positioned as a versatile foundation for the agent era. The model excels in programming, office automation, and long-cycle autonomous execution, with benchmarks showing it leads peers in multiple agent evaluations.

88Z AI editorial1 min read
Flash

Alibaba Cloud Launches New MaaS Model Service Platform 'Qwen Cloud'

At the 2026 Alibaba Cloud Summit, Alibaba Cloud unveiled Qwen Cloud, a new platform designed for AI Agents. It integrates APIs from over 150 mainstream models, including Qwen, DeepSeek, and Kimi. The platform transforms model services into Skills and CLI tools, allowing Agents to automatically handle model selection, invocation, and task execution with a single command, eliminating the need for manual integration code.

88Z AI editorial1 min read