Shanghai AI Lab Releases Open Source 'Shusheng・Wanjuan' 1.0 Multi-Modal Pre-trained Dataset

Translation: The Shanghai AI Lab, in collaboration with the Corpus Data Alliance, has released the "Bookworm・Millions" 1.0 multi-modal pre-training corpus, which includes text, image-text, and video datasets. This open-source corpus exceeds 2TB in total and has undergone fine-grained cleaning and deduplication, featuring diverse integration, meticulous processing, and ease of use with high efficiency. The release of this corpus is expected to promote the application and innovation of large models, and lower the barriers to large model technology.

LobsterAI Launches Image and Video Large Model Matrix, Integrating Four Mainstream Image and Video Generation Models at Once

Domestic AIGC multimodal creation field has made new progress, with the open-source AI product LobsterAI (Lobster) under NetEase Youdao upgraded and officially launched image and video generation capabilities. This upgrade adopts a matrix integration strategy, integrating four mainstream multimodal large models: Seedream, Seedance, HappyHorse, and MiniMax-Hailuo, enhancing creative efficiency and diversity.

Alibaba Cloud BaiLian Fully CLI-Enabled and Open-Sourced: A Single Command to Enable Full Stack Capabilities of AI Agent Orchestration

Alibaba Cloud BaiLian announced on May 29, 2026, that it is fully CLI-enabled and open-sources its CLI project. This move drives a full-stack integration transformation for AI Agent access and development. The CLI encapsulates core capabilities such as mainstream models, workflows, knowledge bases, memory management, web search, and multimodal file processing into a lightweight command-line interface, allowing developers to efficiently use them after installation and authentication.

Xiaomi Launches the Most Powerful Model Series MiMo-V2.5, Official Public Testing Begins

Xiaomi released the MiMo-V2.5 series of large models on April 23 and initiated public testing. The series includes four models, with the core models MiMo-V2.5-Pro and MiMo-V2.5 being open-sourced globally, demonstrating its commitment to promoting an open AI ecosystem. This update is not only a product iteration but also a comprehensive upgrade of the technology foundation, featuring flagship performance that supports a context length of up to one million and complex task processing.

Ant Forest LingBot Open-Source 2.7T Depth Dataset 2 Million Real Samples Covering 6 Cameras

Ant Forest LingBot Technology opens a large-scale RGB-D depth dataset called LingBot-Depth-Dataset, containing 3 million high-quality samples, of which 2 million are collected from real scenes and 1 million are rendered. The total size reaches 2.71 TB, covering 6 mainstream depth cameras. It is currently the largest real-scene RGB-D dataset in the open-source community, providing richer data support for embodied intelligence, spatial perception, and 3D vision fields.

Tapping the Allen Institute for AI: Microsoft Assembles a Super Intelligence Dream Team, Aiming to Reduce Reliance on OpenAI

Microsoft has recruited top research teams from the Allen Institute for Artificial Intelligence and the University of Washington, led by Ali Farhadi, former CEO of Ai2, who joined Microsoft's newly established 'Super Intelligence' department, aiming to strengthen its general artificial intelligence strategy.