Flash

The news, the context, and what changes.

RSS
Flash

Doubao Officially Launches Professional Edition

Doubao Professional Edition is officially launched, introducing a new office task mode based on the Doubao 2.1 series of large models. It supports local computer operations, browser integration, Skills invocation, scheduled tasks, and built-in Office suites, enabling professional work such as document processing, application development, and data analysis. Subscription features three-tier pricing: Standard at 68 yuan/month, Enhanced at 200 yuan/month, and Premium at 500 yuan/month; verified univ

88Z AI editorial1 min read
Flash

Volcengine Launches Doubao Audio Generation Model 1.0

Volcengine has released the Doubao Audio Generation Model 1.0, which supports text or audio reference inputs to generate complete audio end-to-end, including multi-character dialogues, emotions, and background music, while maintaining consistent voice tones over long durations.

88Z AI editorial1 min read
Flash

Anthropic Launches Enterprise AI Collaboration Tool Claude Tag

Anthropic has released Claude Tag, transforming its AI assistant into a virtual colleague within Slack channels. The tool supports task assignment via mentions, automatically breaks down and executes workflows, and features team relay collaboration, organizational memory retention, proactive monitoring alerts, and asynchronous long-task processing to enhance enterprise collaboration efficiency.

88Z AI editorial1 min read
Flash

Qwen Launches Native Language World Model Qwen-AgentWorld

Alibaba Tongyi has launched Qwen-AgentWorld, the first native language world model, supporting agent interaction simulation across seven domains. Trained on tens of millions of real-world trajectories, it outperforms top models like GPT-5.4 and Claude Opus 4.8 in proprietary benchmarks.

88Z AI editorial1 min read
Flash

TRAE Work Launches Design Mode: Chat to Design, Canvas as Prototype

TRAE Work has launched a new Design mode, available simultaneously on desktop and web. Users can generate design proposals through natural language conversations, modify them in real-time on the canvas, and create interactive prototypes. The mode supports importing existing design assets from platforms like Figma and allows one-click export of designs to Code mode, enabling a seamless workflow from design to development.

88Z AI editorial1 min read
Flash

Momenta Targets HK Listing as 'First Physical AI IPO'

Autonomous driving firm Momenta is pursuing a Hong Kong listing, positioning itself as a builder of physical AI foundation models. Its R7 world model is mass-produced, leveraging massive data and a three-layer architecture to understand the physical world. Revenue has tripled in three years, with licensing income surging 42-fold as the business model shifts to scalable licensing fees.

88Z AI editorial1 min read
Flash

Shadow眸 Technology Completes Hundreds of Millions in New Funding Round

Shadow眸 Technology announced the completion of a new funding round worth hundreds of millions, led by Cathay Capital and Shanghai State-owned Capital Venture Guide, with existing shareholders continuing to participate. The company simultaneously released the Hyper3D Rodin Gen-2.5 model, introducing an LLM think-then-generate mechanism with five levels of thinking depth, capable of generating million-face models in as fast as 4 seconds, and launched the world's first native 3D texture model with

88Z AI editorial1 min read
Flash

ByteDance Launches New Agent Model Seed2.1

ByteDance has released the Seed2.1 series of models, positioned as new agents for real-world productivity scenarios. The model shows significant improvements in general agent capabilities, end-to-end code delivery, and multimodal understanding. It leads in multiple benchmarks and surpasses Claude Opus 4.6 in coding crowd-testing win rates.

88Z AI editorial1 min read
Flash

ByteDance Launches Latest Video Generation Model Seedance 2.5

ByteDance officially released the Doubao video generation model Seedance 2.5 at the Volcano Engine FORCE conference. Currently in the final stages of beta testing, the model is scheduled for launch in early July, featuring breakthroughs in video duration, multimodal input, and controllable editing.

88Z AI editorial1 min read
Flash

Baidu Open-Sources End-to-End Long Document OCR Model Unlimited-OCR

Baidu has released the open-source OCR model Unlimited OCR, specializing in long document parsing. It achieved a new state-of-the-art comprehensive score of 93.23% on the OmniDocBench benchmark, surpassing DeepSeek OCR. The core innovation is the Reference Sliding Window Attention mechanism, which uses a soft forgetting strategy to maintain constant KV Cache, ensuring inference speed does not drop as document length increases, with TPS improving by approximately 35% at 6000 tokens.

88Z AI editorial1 min read