Recently, the Modelers community officially launched Step-Video and Step-Audio, two open-source multimodal large models developed by Step-Star. These models are designed for video generation and voice interaction, respectively, aiming to provide developers and enterprise users with more powerful AI tools. Step-Video, formally known as Step-Video-T2V, is a 30-billion parameter model, making it the world's largest open-source video generation model. This model can directly generate 20...

Baidu Research has unveiled BGE-VL, a groundbreaking multimodal vector model poised to revolutionize information retrieval. This advanced model promises significant improvements in search accuracy and efficiency.

On March 6th, the Beijing Academy of Artificial Intelligence (BAAI) announced the open-sourcing of its multimodal vector model, BGE-VL. This achievement marks a significant breakthrough in the field of multimodal retrieval. The BGE-VL model achieves state-of-the-art results on various multimodal retrieval tasks, including image-text retrieval and compositional image retrieval, significantly improving the performance of multimodal retrieval.

Samsung recently announced that the Galaxy S25 series launched in the Chinese market will now integrate the DeepSeek-R1 large language model, significantly enhancing its AI functionality. According to Samsung China, users can access features including in-depth reasoning, internet search, emotional intelligence, contextual memory, and visual analysis without downloading additional apps or registering accounts. Simply update the Samsung Members app to the latest version or access DeepSeek-R1 through Samsung's Smart Home or Bixby's SmartThings.

On March 3rd, Shanghai Secret Tower Network Technology Co., Ltd. announced that its AI search function has added a new "video" search module, further expanding its coverage of multimodal data. This new function, based on the analysis and understanding of hundreds of millions of video contents, helps users find the learning and entertainment video resources they need more efficiently.

DeepSeek launched an official account on Zhihu and published a technical article, "DeepSeek-V3/R1 Inference System Overview," detailing its model inference system optimization and cost-profit margin information for the first time, marking the end of the highly anticipated DeepSeek Open Source Week. The article highlights two key optimization goals for the DeepSeek-V3/R1 inference system: higher throughput and lower latency. To achieve these goals, DeepSeek employs large-scale cross-node expert parallelism (EP) technology, although...

Baidu announced that its latest Ernie large language model, version 4.5, will officially launch on March 16th. Ernie is a significant development in Baidu's AI strategy and has had a profound global impact since its initial release two years ago.

Baidu is preparing to release its next-generation AI model, Wenxin 4.5, expected to launch in mid-March. Internal sources suggest significant improvements in core areas such as reasoning, marking another major advancement in Baidu's AI technology.

Microsoft has unveiled Phi-4, a new multimodal and mini-model designed to significantly improve processing capabilities across speech, vision, and text. This release represents a notable advancement in AI technology.

Microsoft has open-sourced Magma, a multimodal AI agent designed to enhance shopping experiences and improve robotic control. This powerful tool offers new possibilities for interaction and automation.

Microsoft recently open-sourced a new multimodal AI Agent foundation model called "Magma" on its official website. This novel AI boasts the ability to traverse both the digital and physical worlds, processing various data types such as images, videos, and text simultaneously. Unlike traditional AI assistants, Magma's unique capability lies in its psychological prediction function, allowing for more accurate understanding of the intentions and future behavior of individuals or objects within videos. Magma's applications are vast; users can leverage this AI for a wide range of tasks.

On February 25th, at the 2025 Global Developer Pioneer Conference, SenseTime announced a comprehensive upgrade to its AI productivity tool, the Xiao Huan Xiong family. This upgrade significantly enhances its multimodal capabilities, accelerating the application of AI and advancing its evolution towards AI Agents. The improvements not only boost performance but also reaffirm AI's fundamental purpose: to serve as a powerful productivity tool.