Flash

The news, the context, and what changes.

RSS
Flash

Google Unveils Next-Gen World Model Genie 4

Google DeepMind has released Genie 4, a model that generates interactive virtual worlds from text, featuring real-time physics simulation and native audio generation for synchronized sound effects and character dialogue.

88Z AI editorial1 min read
Flash

Ant Group Launches Personal AI Assistant 'Lingguang 2.0' Capable of Managing Affairs

Ant Group has released Lingguang 2.0, an upgraded personal AI assistant with cross-session memory and long-term task scheduling capabilities. Relying on independent cloud hosts, it transitions from an executor to a manager of affairs, proactively handling finance, childcare, chores, and travel. The system integrates the APASS security framework, using separated hosting for authorization keys and Agents to ensure privacy, allowing users to adjust or revoke permissions at any time.

88Z AI editorial1 min read
Flash

Baidu Wenxin Open-Sources End-to-End Document Parsing Model Unlimited OCR 2

Baidu Wenxin has released and open-sourced Unlimited OCR 2, an end-to-end document parsing model capable of continuously recognizing documents exceeding 3,600 pages. Its core SR-SWA attention mechanism, combined with chunked decoding, reduces peak KV cache for hundred-page tests by approximately 96% and increases 16-way concurrent throughput by 3.65 times.

88Z AI editorial1 min read
Flash

Microsoft Launches Image Editing Model MAI-Image-2.7-Preview

Microsoft has released the MAI-Image-2.7-Preview image editing model. AI CEO Suleyman stated that it ranks second globally on the Artificial Analysis leaderboard, trailing only OpenAI's GPT Image 2.5 Sunburst and surpassing Google, Meta, and xAI. As a successor to the MAI-Image-2.6 released in August, this model scores 24 Elo points higher than the next best model in scene and style editing.

88Z AI editorial1 min read
Flash

Anthropic Officially Files for Nasdaq IPO

Anthropic has publicly filed for a Nasdaq IPO under the ticker "ANT," targeting a valuation of up to $2 trillion. Although H1 revenue exceeded $16 billion, significant paper losses were driven by changes in the fair value of financing instruments. Growth is fueled by Claude Code and enterprise clients, while massive long-term compute contracts bring challenges related to high capital expenditure and supply chain dependence.

88Z AI editorial1 min read
Flash

SpaceX Completes $40 Billion Debt Financing to Purchase NVIDIA GPUs

SpaceX CEO Elon Musk announced that the company has successfully completed $40 billion in debt financing, primarily to purchase GPU chips, with delivery expected in the first or second quarter of next year. Capital expenditures for the fourth quarter are projected between $35 billion and $40 billion. Musk noted that companies lacking revenue sources outside AI data centers will face financing difficulties, potentially leading to consolidation or bankruptcies among new cloud providers, though Spa

88Z AI editorial1 min read
Flash

OpenAI Launches Efficient Execution-Oriented Large Model GPT-6.1 Terra

OpenAI has released GPT-6.1 Terra, focusing on low cost, high speed, and AI Agent office scenarios. The model features significant enhancements in visual understanding, computer operation, factual accuracy, and coding, capable of handling blueprints, invoices, desktop software, and research tasks. Its standard API price is 50% lower than the previous generation, and it has been integrated into ChatGPT Work and Codex.

88Z AI editorial1 min read
Flash

Jimeng AI Launches Short Drama Studio

Jimeng AI has launched a Short Drama Studio, integrating self-developed models to cover the entire workflow from script generation and art design to storyboard preview and post-production. It supports long-text analysis, 3D director preview, and voice reconstruction, while facilitating global localization through remake-style image translation.

88Z AI editorial1 min read
Flash

Tencent Hunyuan Open-Sources Simultaneous Interpretation Model Hy-SimulMT

Tencent Hunyuan has open-sourced Hy-SimulMT, a simultaneous interpretation model supporting real-time Chinese-English translation and voice cloning. With a response time of about 3 seconds, it is faster than human interpreters and achieves expert-level quality, outperforming mainstream open-source models and many commercial APIs.

88Z AI editorial1 min read
Flash

Moonshot AI Launches Latest Flagship Model Kimi K3.1 Pro

Moonshot AI has officially released Kimi K3.1 Pro, the first model in the K3.1 series, featuring 3 trillion parameters and support for a million-token context. It significantly enhances capabilities in coding, 3D creation, financial research, and long-horizon agents, while introducing a high-speed version that is approximately two to three times faster. The model is now available via the official website, App, Work, Code, and API, with full weights scheduled for release before November 4.

88Z AI editorial1 min read
Flash

Tencent Hunyuan Open-Sources Interactive Video World Generation Model Hy-Nomas

Tencent Hunyuan has released Hy-Nomas, an open-source interactive video world generation model. It supports text and camera trajectory control while maintaining long-range consistency. Optimized for local inference on a single RTX 4090 with LoRA fine-tuning support, it is suitable for gaming, simulation, and film pre-visualization.

88Z AI editorial1 min read