Search AI Products and News

NVIDIA officially released Llama Nemotron Nano VL on June 3, 2025. This compact visual-language model (VLM) is specially optimized for document intelligence processing. The model topped the OCRBench v2 benchmark test, showcasing its outstanding capabilities in handling complex documents, charts, and video frames. With efficient inference performance and flexible deployment options, Llama Nemotron Nano VL provides enterprises with solutions from the cloud.

The official release of Tencent HunYuan 3D generative model version 2.5 marks a significant leap into the ultra-high-definition era of 3D generation technology. This upgrade not only achieves a qualitative leap in modeling precision but also provides users with a more efficient and convenient creative experience, further lowering the barrier to entry for 3D content creation.

Hugging Face recently released its top model leaderboard for the second week of April 2025, encompassing various modalities such as text generation, image generation, and video generation. This highlights the rapid iteration and diverse applications of AI technology. According to AIbase, the models featured in this week's leaderboard not only showcase the innovative vitality of the open-source community but also reflect technical trends such as low-precision training and multi-modal generation. Below is an analysis of the leaderboard highlights, with professional insights provided by the AIbase editorial team. Text Generation Models: Balancing Efficiency and Specialization

Today, on the second anniversary of Baichuan Intelligent's founding, the company's founder and CEO Wang Xiaochuan released an all-staff letter, reviewing the past two years' achievements and shortcomings, and clarifying the company's current and future development direction. Wang Xiaochuan emphasized that Baichuan Intelligent will firmly pursue its technological vision and business direction in the medical field, promoting the deep application of AI technology in the medical industry through the path of "creating doctors - changing paths - promoting medicine".

Recently, Professor Zhai Jidong's team from the Institute of High Performance Computing at Tsinghua University, in collaboration with Tsinghua-affiliated innovative enterprise, Qingcheng Extreme Intelligence, announced the open-sourcing of Chitu, a groundbreaking large model inference engine. This innovative technology marks another significant breakthrough for China in the AI field, particularly in inference engine development. A core highlight of the Chitu engine is its ability to natively run FP8 precision models on non-Nvidia Hopper architecture GPUs and various domestic chips.

A German research team has developed and released FIORA, an open-source graph neural network (GNN) model designed to improve the accuracy of mass spectrometry analysis. This advancement is expected to significantly contribute to the progress of precision medicine.

The University of Science and Technology of China (USTC) recently announced a significant breakthrough. Professor Sun Cheng's research group and collaborators have successfully developed the "TIMES" scoring system for predicting liver cancer recurrence, and have translated it into a publicly available, free AI diagnostic tool. The findings were published in the international academic journal Nature on March 13th, Beijing time. Liver cancer, the third leading cause of cancer-related deaths globally, has a recurrence rate of up to 70% after surgical resection. Accurate prediction of liver cancer recurrence has long been a medical challenge. The spatial distribution of immune cells in the tumor microenvironment significantly impacts patient prognosis, but previously...

A recent study suggests that artificial intelligence (AI) has the potential to surpass human intelligence in the future. The research delves into the complexities of AI and the human brain and proposes a new theoretical framework, indicating that by applying AI technology at the cellular level of neuroscience, AI could infinitely approach the functionality of the human brain, potentially even surpassing human intelligence levels. The human brain is widely regarded as one of the most complex systems in the universe, while artificial intelligence is seen as one of the most important technologies in human history. A core question looms: Will artificial intelligence ultimately surpass human intelligence?

Recently, US defense contractor Anduril Industries announced the launch of two new autonomous aircraft, Bolt and Bolt-M. These drones are equipped with advanced artificial intelligence technology and are touted as easy to operate and highly precise, aiming to change the traditional use of drones. The Bolt is a reconnaissance drone with a maximum flight range of over 20 kilometers, a flight time of over 45 minutes, and weighs approximately 5.4 kilograms. Anduril states that this drone is not only suitable for reconnaissance but can also

The computer science research team at the University of Idaho's Coeur d’Alene campus has developed an autonomous robot that uses artificial intelligence for precision weeding, capable of identifying and targeting half-inch wide weeds through electric shock eradication. After demonstration at the Coeur d’Alene nursery conducted by the Forestry Bureau, plans are underway to improve the hardware and electrical components to enhance the AI's accuracy in weed detection. This project, in collaboration with the USDA, aims to increase reforestation efficiency in the national forest system, reduce labor costs and pesticide use, and bolster local ecosystems.

NVIDIA has released StormCast, a generative AI model designed for high-fidelity atmospheric dynamics simulations, aiming to accurately predict mesoscale weather, which is crucial for disaster planning and mitigation. This model made its debut in papers co-authored with research institutions, providing groundbreaking predictive capabilities for frequently occurring extreme weather events, such as the over $150 billion annual losses in the U.S. StormCast is part of the NVIDIA Earth-2 digital twin cloud platform, integrating AI, physics simulations, and computer graphics.

The Institute of Automation of the Chinese Academy of Sciences, in collaboration with Tsinghua University, Peking University, and other institutions, has proposed a method for constructing brain-like neuron models based on endogenous complexity. This method demonstrates the equivalence of the dynamic characteristics of the LIF model and HH model in spiking neural networks and enhances the endogenous complexity of the computational unit through microarchitecture design, allowing the HH network model to simulate larger-scale characteristics of the LIF network model. The research team has also simplified the model to the s-LIF2HH model, and validated its ability to capture complex dynamic behaviors through simulation experiments.

ControlNeXt, developed by the Jia Jiaya team, is an open-source image/video generation guidance tool designed to simplify the use of AI models. This tool is lightweight and compatible with models from the Stable Diffusion family, such as SDXL and SD1.5, supporting various control modes, including edge guidance, pose control, masking, and depth of field control. Its 'weight loss secret' lies in adopting a more efficient structure, replacing the 'big eater' control branch with a 'light meal package' composed of a small number of ResNet blocks, reducing the model size.