Search AI Products and News

DeepMind's Mixture-of-Recursions (MoR) enhances model efficiency via dynamic token routing and recursive depth allocation, outperforming Transformers with fewer parameters. Its selective caching reduces memory pressure, proving especially effective above 360M scale, offering optimized AI deployment solutions.....

OpenMed released 380+ free medical NER models on Hugging Face (Apache2.0). These models (109M-568M params) rival paid alternatives, integrated into major AI ecosystems. Addressing global healthcare shortages, the project allows free use/modification, with COVID screening API previously developed in 5 days. Team plans to expand model library for open-source medical AI.....

A Chinese team has released the open-source AI operating system NeuralOS, which realizes the GUI concept for the AI era proposed by Karpathy. The system uses two modules, RNN and neural renderer, to predict and simulate the Windows operating interface in real-time, accurately displaying user operation feedback. The development team trained the system using a large number of operation videos, and it can currently accurately predict user operations, but still has limitations in handling fast keyboard input. NeuralOS already offers an online trial version, showcasing a new experience of an AI-generated operating system. As the code...

Mistral releases open-source audio model Voxtral, offering cost-effective AI solutions with multilingual support. Three versions available, starting at $0.001/min. Outperforms Whisper at lower cost.....

Intel's former CEO Pat Gelsinger collaborates with Gloo to launch the Flourishing AI (FAI) benchmark, evaluating AI models' alignment with human values. Based on Harvard's prosperity research, it covers 6 dimensions including virtues, relationships, and happiness, adding a unique 'faith & spirituality' category. The benchmark aims to guide AI development toward human well-being.....

At the 2025 World Artificial Intelligence Conference, it was revealed that Shanghai has filed 82 large models and is actively promoting AI demonstration applications in key industries such as manufacturing and finance. Xuhui Moshu Space and Pudong Moli Community have become industrial carriers, gathering 500 and 200 AI companies respectively. Shanghai has established a full-cycle financing support system from the early stages to the mature stage through national and municipal artificial intelligence funds, with a focus on key areas such as computing power and language data.

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.....

Aliyun open-sources the network agent WebSailor. Its 32B and 72B versions performed well in the BrowseComp evaluation, surpassing multiple closed-source models, ranking just behind OpenAI DeepResearch. The project has been released on GitHub with construction plans and datasets, promoting open innovation in the AI field and providing developers with a smarter web interaction tool.

In August 2024, iFlytek officially launched the Starfire Ultra Human-like Interactive Technology. Through end-to-end speech modeling and multi-dimensional emotional disentanglement training, it achieves three core breakthroughs: response speed, emotional resonance, and controllable speech expression. This technology can accurately detect emotional fluctuations in user speech and respond with appropriate tone in real time, while supporting dynamic adjustment of speech rate, voice, and character settings. It marks a significant leap from 'functional implementation' to 'emotional connection' in voice interaction. Currently, the Super Human-like Interactive API has been officially launched on the iFlytek Open Platform, allowing developers to access the technology at a low cost.

Musk announces Grok4 AI model launch on July 11, sparking debates on AI ethics due to controversial 'politically incorrect' responses. New rules require multi-source analysis and evidence-based disputes.....

OpenAI plans to launch GPT-5 this summer, integrating multiple model capabilities. The new version combines reasoning of 'O-series' and GPT's multimodal advantages for enhanced performance. It aims to simplify user experience by eliminating model switching. GPT-5 will boost functionality and usability, though exact release timing remains unclear.....

China's smart speaker market rebounded in 618 2025, with online sales reaching 802K units (+7.5% YoY) and revenue up 15.2%. AI model adoption (36.8% products) drove growth. Xiaomi and Baidu's new models led innovation in voice interaction. The market revival after 4-year slump signals sustained AI-driven expansion.....

Recently, a research team from Renmin University, Shanghai Artificial Intelligence Laboratory, University College London, and Dalian University of Technology revealed an important finding in the reasoning process of large models: when the model is thinking, the 'thinking words' it uses actually reflect a significant increase in its internal information. This research result provides a new perspective for better understanding the reasoning mechanisms of artificial intelligence through methods of information theory. You may have seen large models output some language that seems human-like when answering questions, such as "Hmm..." or "Let me think...".

In the field of artificial intelligence, reward models are a critical component for aligning large language models (LLMs) with human feedback, but existing models face the issue of "reward hacking." These models often focus on superficial features, such as the length or format of responses, rather than identifying genuine quality metrics, such as factual accuracy and relevance. The root cause lies in standard training objectives failing to distinguish between spurious associations and true causal drivers present in the training data. This failure leads to fragile reward models (RMs), which generate misaligned policies.

According to KPMG China's recent report, "The First 50 Health Tech Companies," China accounts for more than 70% of the global release volume of medical large models. This data not only demonstrates China's rapid development in the field of intelligent healthcare, but also reflects the wide application of large language models in the healthcare industry. The report points out that about 65% of the currently released medical large models are large language models. These models can process and generate natural language, playing a significant supporting role in the analysis of medical data, patient communication, and scientific research.

In the long-standing copyright infringement lawsuit filed by The New York Times against OpenAI, the case has made significant progress. According to Ars Technica, the federal judge presiding over the case has authorized The New York Times and its co-plaintiffs, The New York Daily News and the Investigative Reporting Center, to access OpenAI's user logs, including deleted content, to accurately determine the scope of the infringement. The New York Times believes that ChatGPT users may delete their history after bypassing the paywall, and therefore it is necessary to conduct large-scale data collection.