Search AI Products and News

Ali International's AI solution Marco won the SAIL Award at the 2025 World Artificial Intelligence Conference, becoming the first product in the cross-border e-commerce field to receive this honor. The solution has been integrated with several leading cross-border service providers such as Miaoshou and Dianxiaomi through technology sharing, helping merchants improve operational efficiency. In 2025, the number of AI API calls from external ecological partners surged 23 times, indicating a strong market demand for intelligent services. With the development of cross-border e-commerce, the application of AI in areas such as data analysis, consumer forecasting, and supply chain management is becoming increasingly widespread.

VLM2Vec-V2, a multimodal embedding framework developed by Salesforce Research and other institutions, breaks through the limitations of traditional models. Based on the Qwen2-VL architecture, this framework innovatively unifies image, video, and visual document retrieval tasks. It adds five new evaluation tasks to expand the MMEB dataset. With key technologies such as dynamic resolution and M-RoPE, it achieved an average score of 58.0 across 78 datasets, leading in performance, especially in video tasks. Although its document retrieval is slightly behind ColPali, it

Stanford University opensourced the AI tool OctoTools, integrating 11 professional tools to address complex reasoning tasks. The tool adopts a standardized tool card design, including functional modules such as image recognition and mathematical calculations. It follows a three-stage workflow: the planner analyzes the task, the executor operates the tool, and the validator ensures the results. Testing shows its outstanding performance in 16 benchmark tests, enabling efficient handling of cross-disciplinary issues such as mathematics and science, significantly improving the reliability and multi-step reasoning capabilities of AI systems. Open-source repository: github.com/oct

Moonshot AI officially released its latest creation - the Kimi K2 model, and simultaneously announced its open source. This foundation model based on the MoE architecture has gained widespread attention in the AI field since its release, thanks to its strong coding capabilities and excellent general Agent task processing abilities. The Kimi K2 model has a total of 1T parameters, with 32B activated parameters. It has achieved top performance among open-source models in a series of benchmark performance tests such as SWE Bench Verified, Tau2, and AceBench.

Kunlun releases Skywork-R1V3.0, an open-source multimodal model achieving expert-level reasoning. It excels in physics (52.8), logic (59.7), and math (77.1), outperforming Claude-3.7-Sonnet and GPT-4.5 with only 25K training samples.....

In the digital age, efficient note-taking tools have become an essential part of knowledge management. Recently, a cross-platform AI note-taking software called NoteGen has quickly gained popularity. It supports five major platforms: Windows, MacOS, Linux, iOS, and Android, and offers free multi-device data synchronization. With native Markdown formatting and strong integration with third-party large models, it redefines the note-taking experience. Full platform support and free synchronization seamlessly connect NoteGen, thanks to its powerful cross-platform compatibility.

On June 29, 2025, the Alibaba International AI Team officially released the new multi-modal large model **Ovis-U1**, marking another major breakthrough in the field of multi-modal artificial intelligence. As the latest masterpiece of the Ovis series, Ovis-U1 integrates multi-modal understanding, image generation, and image editing functions, demonstrating powerful cross-modal processing capabilities, providing new possibilities for developers, researchers, and industry applications. This is a detailed report on Ovis-U1 by AIbase. Ovis-U1

Google DeepMind recently published a blog post, officially introducing the new Gemini Robotics On-Device localized robot AI model. The model uses a vision-language-action (VLA) architecture, enabling precise control of physical robots without cloud support. The biggest highlight of the new model is that it runs entirely locally on the robot device, achieving low-latency response capabilities. This feature makes it particularly suitable for environments with unstable network connections, such as critical application scenarios like medical facilities.

Recently, a new general intelligent agent called MiniMax Agent has been officially launched, aiming to provide efficient solutions for complex long-term tasks and has attracted widespread attention in the industry. MiniMax Agent: A breakthrough work of general intelligent agents - MiniMax Agent is an intelligent agent designed specifically to solve long-term complex tasks, featuring expert-level multi-step planning capabilities, flexible task decomposition mechanisms, and end-to-end execution efficiency. Its design concept is to make AI a reliable teammate.

A comprehensive analysis of MCP Stack's open-source marketing cloud platform functions, detailing its technical advantages in multi-channel integration, AI automated marketing, real-time data analysis, and providing implementation suggestions for enterprise-level marketing automation solutions.

In recent years, the field of artificial intelligence has undergone tremendous changes, especially with large language models (LLMs) making remarkable progress in multi-modal tasks. These models have demonstrated strong potential in their ability to understand and generate language, but most current multi-modal models still adopt auto-regressive (AR) architectures, which limit inference processes to be rather monotonous and lacking flexibility. To address this limitation, a research team from The University of Hong Kong and Huawei Noah's Ark Lab has proposed a brand new model – FUDOKI, aiming to break these constraints. The core innovation of FUDOKI is

A new study led by Apple researchers has poured cold water on the current large reasoning models (LRMs) that are highly anticipated. The research found that when solving complex tasks, reasoning models designed to simulate thinking processes, such as Claude3.7Thinking and Deepseek-R1, not only failed to show advantages but also exhibited serious problems like insufficient reasoning and performance collapse. This study tested four classic logical puzzles: the Tower of Hanoi, Checkers, River Crossing, and Block World. These puzzles allow precise control over the tasks.

The 2025 Meituan Accommodation Ecosystem Partner Conference was recently held in Qingdao, Shandong. Meituan Travel released the first domestic AI tool for hotel merchants - Meituan Ji Bai. Wang Puzhong, CEO of Meituan Core Local Business, stated that Meituan is stimulating new accommodation demand through cross-scenario integration and ecological collaboration, helping hotel merchants unlock greater growth opportunities. Data shows that the number of overnight stays for out-of-town travel accommodations increased by 14% year-over-year, while cross-sector demands such as accommodation + catering surged nearly 87%, accommodation + entertainment grew by 99%, and accommodation + air tickets increased by 36%. During this year's May Day holiday, Meituan...

In today's modern workplace, the importance of IT support cannot be overstated. Whenever employees encounter issues such as locked computers or inaccessible applications, they always hope to contact IT support promptly. However, help desk staff are often busy handling other requests, which may result in employees not receiving timely support. Image source note: The image is generated by AI and provided by Midjourney. Andrei Serban was once the product leader of the application and integration team at Rippling, and he...