GitHub

Welcome to the "Ai Daily" section. This is your daily guide for exploring artificially intelligent worlds, and we give you the hot spots in the field of AI every day, focusing on developers, and helping you to understand technological trends and innovative AI applications.

Fresh AI products click on: https://app.aibase.com/zh

1 byte beat open source Lance 3B: Handle visual understanding and generation with a brain

The byte beat open source of its original unified polymodular model Lance, which achieves full functional coverage with 3B parameters, breaks the technical barriers between the understanding model and the generation model. Lance has harmonized image, video understanding, generation and cross-modular editing by sharing context and capability decoupling.

[AiBase Summary:]

Lance uses a parallel design with shared context and capability to harmonize multi-modular tasks.

🚀 3B parameter mass achieves full functional coverage, breaking the technical walls of traditional models.

• Open-source Apache 2.0 agreement, which allows civilian-level computing to operate and reduces deployment costs.

2. Distribution of the GLM-51 high-speed version: 400 tokens/s worldwideAPINew limit

The GLM-51 high-speed version of the API was released with a view to updating the API ceiling of the global large model at a speed of 400 tokens/s, achieving the full size of the flag class with extremely low delays, and promoting the efficient development of AI applications through system-level engineering to optimize model performance.

[AiBase Summary:]

• The GLM-51 high-speed version of the API of the smart spectrum achieves 400 tokens/s output speeds, updating the API ceiling of the large global model.

• Achieving full-scale flag-size capability with extremely low delays and breaking industry practices.

Enhancement of model performance through system-level engineering optimization, including synergistic optimization of reasoning engines, movement control systems and infrastructure layers.

3, CapCut and Gemini. Collaborative in-depth integration: AI Creative tool for intelligent interconnection

CapCut and Google Gemini App collaborates that users can use CapCut 's advanced creative and editorial functions directly within the Gemini application to further promote the spread and innovation of the AI tool in the field of content creation.

[AiBase Summary:]

CapCut and Go.The ogle Gemini App partnership allows users to directly call CapCut 's advanced creative and editorial functions within the Gemini application.

The aim of this collaboration is to create a more seamless and efficient AI creative experience and to reduce the cost of cross-application switching.

CapCut indicates that future creative methods will be more interactive, visualized and intelligently integrated.

4.OpenAI Release ChatGPT For PowerPoint: Generate a PPT and take the initiative, Bug

OpenAI Launch ChatGPT For PowerPoint plugins, which enable users to quickly generate and optimize PPT content through simple commands, with smart analysis and modifications, greatly enhance office efficiency.

[AiBase Summary:]

The zero threshold is free and the ChatGPT for PowerPoint plugin is available to users worldwide.

💡 Supports the creation of new PPTs from zero, one-key changes/colouring of pages, or even " doubledisk " programmes.

• Introduction of key operational validation mechanisms to ensure that each change is controlled.

Five, WordPress 7.0 Official launch: original integration AI into a new era of smart stations

WordPress 7.0 official launch, original IAI capability, marks the beginning of the intellectualization of the web site. The new edition has been overhauled in content creation, back-office interface and mobile-end experience, bringing more efficient and fluid site and editorial experiences to users.

[AiBase Summary:]

Primary IAI capacity to improve content creation efficiency.

• Modernization of the back-office interface to optimize user experience.

📱 Mobile-end self-defined enhancements to enhance responsiveness to editing.

Six, Spotify, join hands with Globe.

Spotify partnered with Globe Music in the introduction of the al-tone and mixer functions, marking a major change in the area of music copyright. This function is based on legal authorization, provides users with entirely new ways of creating and safeguards the interests of artists through a reasonable system of division. This initiative not only enhances the market competitiveness of Spotify, but also poses a strong challenge to other AI music platforms.

[AiBase Summary:]

Spotify and Globe Music reached an Ai-Ronging and Mixing Agreement, which provides fans with a legitimate creative tool.

• Underline the golden principles of “consensual consent, honour and fair remuneration” as distinct from the violation patterns of other AI platforms.

📈Spotify stock price increased by 13% due to the AI strategy, demonstrating its strong influence in music copyright.

Uniclaw2026 Common: AI is entering the multi-person social model

The brand-new AI live-in communication product Uniclaw, launched by the Beijing-based Vientiane Artificial Intelligence Technology Ltd., has broken the traditional single-talking dialog model, upgraded AI from a single-person efficiency tool to a social Agent for group collaboration and launched the AI multi-person collaboration model. The article also details the three core roles of UniClaw and the open Agent application community.

[AiBase Summary:]

✨UniClaw as an AI live communication product breaks the traditional single-talking dialog and opens the AI multi-person collaboration model.

The AI Smart Body (Agent) assumes the role of information hub, communications lubricant and active focal point in the cluster to enhance the efficiency of collaboration.

• The open Agent application community lowers the threshold for use, and the user can introduce a specific feature, Agent, with one key, to be inserted.

No, no, no, no.

LongCat-Video-Avatar1.5, the official open-source commercial digital-generated model of the United States Dragon Cat Group, achieves an overall leap forward in lip synchronisation, physical reasonableness, long video stability and significantly enhances the commercial application value and user experience of the model through several technological upgrades.

[AiBase Summary:]

🧠 The model upgrades the audio feature extraction encoder from Wav2Vec2 to Whisper-large to enhance the capture capacity for acoustic changes and pronunciation.

Introduction of GRPO technology to optimize the alignment of hands with continuity and to address hand malformations and incoherence.

Using DMD technology, reasoning efficiency is increased 15 times, and generating a 10-second video takes about 1 minute.

Further links: https://github.com/metuan-longcat/LongCat-Video