Flash

The news, the context, and what changes.

RSS
Flash

HiDream Officially Launches Content Creation Agent vivago R1 'Gou Da'

HiDream's content creation agent vivago R1 has launched globally, with its domestic version 'Gou Da' simultaneously upgraded. As the world's first infinite-duration creation agent, it supports one-time output of 5-minute high-quality videos with continuous extension capabilities, achieving an 85% effective usability rate. The product adopts a chat-first mode, driven by both foundation models and agents to automatically decompose scripts, storyboards, and sound effects.

88Z AI editorial1 min read
Flash

OpenAI Launches New Image Generation Model ChatGPT Images 2.5

OpenAI has released ChatGPT Images 2.5, significantly improving detail, editing precision, and speed, with a 50% reduction in latency. New features like sketches and templates are now available, along with two new API models, Flare and Sunburst.

88Z AI editorial1 min read
Flash

Xiaomi Opens Invitational Testing for Xiaomi MiMo Desktop

Xiaomi has launched invitational testing for MiMo Desktop, entering the desktop Agent market to compete with Tencent, Alibaba, and ByteDance. Users can provide feedback via real tasks and access MiMo-X-Pro-Preview and MiMo-X-Flash-Preview for free on a limited basis.

88Z AI editorial1 min read
Flash

Meta Launches First Personal AI Agent Muse

Meta has launched Muse, its first personal AI agent capable of autonomously performing tasks such as sending emails, selling cars, and booking travel on behalf of users. The product is now available in the United States with a subscription fee. Muse can access sensitive data including emails, calendars, payments, and health records. Users can choose which apps to connect and retain the right to revoke access at any time.

88Z AI editorial1 min read
Flash

Ant Group Open-Sources Unified Image Generation and Editing Model LLaDA-Image

Ant Group's InclusionAI has open-sourced LLaDA-Image, a 6B-parameter text-to-image model. Featuring self-conditioned learning, unified generation and editing, and TwinFlow distillation that reduces steps from 50 to 2–4, it ranks first among open-source models in both Chinese and English on Qwen-Image-Bench, with a total score between GPT Image 1 and Imagen 4.0 Ultra.

88Z AI editorial1 min read
Flash

DeepSeek Tests New Model DeepSeek V4.1 Flash

DeepSeek has announced the internal testing of the intermediate version of DeepSeek V4.1 Flash. Featuring a new architecture with native multimodal support, the model offers enhanced capabilities, faster speed, and lower costs. Developers can access it by updating the model name while keeping the original base URL. Billing matches DeepSeek V4 Flash, with a limit of 20 concurrent requests per account. The model will automatically expire on September 10.

88Z AI editorial1 min read
Flash

Mistral AI Completes €3 Billion Series D Funding

French AI startup Mistral AI announced the completion of a €3 billion Series D funding round, reaching a post-money valuation of over €21 billion. Led by Samsung Electronics and others, the funds will support advanced research.

88Z AI editorial1 min read
Flash

ModelBest Open-Sources Edge Language Base Model MiniCPM5-2B

ModelBest and OpenBMB have released MiniCPM5-2B, an edge base model that ranks first among open-source models under 4B parameters on the AA leaderboard. It achieved a comprehensive score of 23 and an Agentic Index of 20. The project also open-sourced UltraData, Meshy framework, and JustRL II, with Day0 adaptation for Intel, Rockchip, and Arm.

88Z AI editorial1 min read
Flash

Tencent Hunyuan Upgrades Hy4 Preview Model

Tencent Hunyuan announces an upgrade to the Hy4 preview model, focusing on optimizing long reasoning and excessive self-verification in complex tasks. The new version is fully released after dual monitoring via benchmark metrics and human evaluation. It significantly reduces task rounds and token consumption while maintaining task performance, thereby improving overall efficiency.

88Z AI editorial1 min read
Flash

Musk's SpaceXAI Expands Grok Bot to iPad and Android

Musk's xAI has expanded the Grok Bot platform to iPad and Android systems, achieving coverage across Mac, iPhone, iPad, and Android devices. The platform supports 24/7 collaborative teamwork among AI agents, featuring cross-application login, context retention, and information sharing among multiple bots.

88Z AI editorial1 min read