Flash

The news, the context, and what changes.

RSS
Flash

Meituan LongCat Launches Native Multimodal Model LongCat-2.5

Meituan LongCat has released LongCat-2.5, a native multimodal model with 1.6 trillion total parameters and approximately 48 billion activated per token. Designed for long-horizon agent tasks, it enhances visual understanding and agent programming through long-sequence training and large-scale Agentic RL to improve complex code development and project building.

88Z AI editorial1 min read
Flash

Agibot Launches Embodied Intelligence Model GO-3 Preview

Agibot has released the GO-3 Preview embodied intelligence model, achieving scalable learning across pre-training, post-training, and deployment. The model secured first place in RoboDojo, RoboTwin, and RoboColiseum benchmarks. It uses WRAM to absorb human first-person data during pre-training, employs Genie Evolver for mixed training with real robots, simulation, and world models, and leverages Genie Master to accumulate experience in real-world settings like convenience stores.

88Z AI editorial1 min read
Flash

OpenAI Announces Shared Quota for ChatGPT and Codex

OpenAI announced that starting in December, ChatGPT and Codex quotas for Plus, Pro, and Business plans will be merged and measured by tokens, with only Instant quick chat excluded from the shared pool. This move addresses recent user dissatisfaction after quota increases for some Pro and Business plans did not include Plus users. Meanwhile, Codex has tightened its multi-account policy, potentially restricting logins for devices linked to too many paid accounts. Specific quotas and exact effectiv

88Z AI editorial1 min read
Flash

Kuaishou's Kling AI Completes Approximately 19.8 Billion Yuan Capital Increase

Kuaishou's Kling AI has completed a capital increase of approximately 19.8 billion yuan, setting a record for the largest single financing in the AI video sector. Investors include Alibaba, Tencent, Baidu, and several state-owned entities. Post-closing, controlling shareholder Lucky Labs holds 66.68%, while CEO Gai Kun retains 23.72% voting rights through special share arrangements.

88Z AI editorial1 min read
Flash

NVIDIA Open-Sources Nemotron 3.5 Transcribe Speech Model

NVIDIA has released the open-source speech transcription model Nemotron 3.5 Transcribe. Built on a MoE architecture, it supports multilingual transcription, translation, and speaker diarization, ranking among the top open-source models in various benchmarks.

88Z AI editorial1 min read
Flash

Hangzhou Invests Hundreds of Millions in 01.AI

01.AI has reached a strategic cooperation agreement with Hangzhou City and Linping District, receiving hundreds of millions in combined investment from state-owned assets at both municipal and district levels. With its headquarters established in Linping, the company will jointly build an industrial AI transformation base to drive the deep application of artificial intelligence in core business scenarios such as operational decision-making, production, and supply chain collaboration.

88Z AI editorial1 min read
Flash

LibTV's Three Flagship Models Available with Unlimited Zero-Point Usage for Up to 30 Days

LibTV has launched a Double 11 promotion for AI video creation, offering unlimited zero-point usage of its three flagship models, Wan 3, Kling 4.0, and Minimax H3, for up to 30 days. Seedance 2.5 720P is priced as low as 0.17 yuan per second. Annual members can save 1300 yuan with red packets, receive an additional three months of membership and over 170,000 points, while the Supreme 92K annual package sees a significant price reduction. The creator growth program has also been upgraded, with in

88Z AI editorial1 min read
Flash

Google DeepMind Launches Latest Flagship Large Model Gemini 4 Carbon

Google DeepMind has released its flagship model, Gemini 4 Carbon, which topped AI intelligence benchmarks and surpassed key competitors. The model features long-term processing capabilities for large codebases, focusing on security vulnerability detection and repair, and is currently available only to select cybersecurity professionals.

88Z AI editorial1 min read
Flash

ModelBest Open-Sources MiniCPM-OCR Series

ModelBest open-sources the MiniCPM-OCR series, with the 1.3B model achieving SOTA on two benchmarks and the 2B model being the industry's first OCR result discrimination model.

88Z AI editorial1 min read