Flash

The news, the context, and what changes.

RSS
Flash

Li Fei-Fei's World Labs Launches Atlas, the World's First Multimodal World Model

World Labs, founded by Li Fei-Fei, has unveiled Atlas, the world's first multimodal world model. Built on a multimodal autoregressive diffusion Transformer architecture, it features pixel-level camera control, 3D reconstruction, and spatiotemporal simulation. Outperforming current state-of-the-art models, Atlas is now available for early access to select partners.

88Z AI editorial1 min read
Flash

Meta Launches First Real-Time Audio Perception Model Muse Voice Transcribe

Meta has released Muse Voice Transcribe, its first real-time audio perception model, integrating streaming speech recognition, speaker diarization, and endpoint detection. Supporting over 70 languages and multi-track transcription for more than 20 speakers, the model uses an adaptive latency mechanism to balance speed and accuracy, ranking first on the Artificial Analysis streaming speech-to-text leaderboard.

88Z AI editorial1 min read
Flash

Alibaba Cloud's Enterprise Agent Collaboration Platform 'Wanwu Wujie' Opens for Public Beta

Alibaba Cloud has launched the public beta of 'Wanwu Wujie,' an enterprise-grade agent collaboration platform supporting the Qwen3.8-Max flagship model. The platform breaks down complex tasks among agents with different roles, facilitating work through project spaces, group chats, and asset libraries. It features two main agents, 'Xiao Wan' assistant and 'Xiao You' project manager, with capabilities for organizational memory ownership, permission control, and heterogeneous agent integration, ser

88Z AI editorial1 min read
Flash

Tencent Hunyuan Open-Sources Hy4 Quantized Compression Model 'Hy4 Preview Lite'

Tencent Hunyuan has released the Hy4 preview lite version, utilizing its self-developed Sherry sparse ternary quantization algorithm and MIX-STQ1_0 mixed-precision layer-wise quantization strategy to compress model weights from 1.5 TB to approximately 214 GB. This quantized version performs close to the original BF16 model in tasks such as long-text understanding, multi-turn context retrieval, and coding assistance.

88Z AI editorial1 min read
Flash

DeepSeek-V4-Flash-Vision-Exp Model Officially Open-Sourced

The first multimodal model in the DeepSeek V4 series, DeepSeek-V4-Flash-Vision-Exp, has been open-sourced under the MIT license. It natively supports image inputs in JPEG, PNG, GIF, and WebP formats, offering capabilities such as image description, text recognition, and chart analysis. The model shows significant improvement in visual Agent benchmarks, with multimodal Agent capabilities approaching Opus-4.8, while pure text task performance matches the official V4-Flash release.

88Z AI editorial1 min read
Flash

VAST Completes Series B and B+ Funding Rounds, Totaling Approximately 3 Billion Yuan

AI 3D company VAST has recently completed its Series B and B+ funding rounds, raising a combined total of approximately 3 billion yuan. The round was led by Matrix Partners China, with participation from industrial investors such as Perfect World, BlueFocus, and 37 Interactive Entertainment, as well as financial investors including CDH Investments, CICC, and CMC Capital. Existing shareholders also made oversubscribed follow-on investments. This brings VAST's cumulative financing to about 5 billi

88Z AI editorial1 min read
Flash

Loopit Open-Sources Real-Time Interactive Video World Model Zing-0.5

Loopit has open-sourced Zing-0.5, a real-time interactive video world model based on Wan2.2-TI2V-5B with 5 billion parameters. It supports combined keyboard and text control, generates video at over 24 FPS on a single RTX 5090, and ranks first in the WBench benchmark for real-time world models.

88Z AI editorial1 min read
Flash

Ant Group's Bailin Launches First Finance-Enhanced Model Ling-3.0-flash-Fin

Ant Group has released Ling-3.0-flash-Fin, its first finance-enhanced model. Built on a 124B total parameter and 5.1B activated parameter architecture, it focuses on improving retrieval, reasoning, modeling, and report writing in financial workflows, demonstrating strong performance in various financial benchmarks.

88Z AI editorial1 min read
Flash

a16z Launches $1.1 Billion Machine Age Fund to Bet on AI Hardware

a16z has announced the launch of the $1.1 billion Machine Age Fund, focusing on AI physical infrastructure including chips, memory, networking, data centers, robotics, and home AI devices. The fund notes that as AI shifts from chat to reasoning and coding, compute demand is growing exponentially, challenging existing supply chains and physical limits, necessitating restructuring.

88Z AI editorial1 min read
Flash

OpenAI Announces Termination of Partnership with Cursor

OpenAI has announced the termination of its partnership with Cursor following its acquisition by SpaceX, with direct model access ending on November 12. Citing compliance concerns and the security needs of the upcoming Astra model, OpenAI is tightening distribution controls.

88Z AI editorial1 min read