Recently, the latest artificial intelligence model Grok3 launched by xAI quickly climbed to the top of the App Store in the United States, becoming a hot topic among users. Behind this achievement, in addition to Grok3's powerful features, are the outstanding contributions of two Chinese scholars. They are Assistant Professor Jimmy Ba from the University of Toronto and Tony Wu, co-founder of xAI. It is worth noting that Tony Wu is not only the only post-95 scholar in the team, but his academic background and development experience are also impressive.

In the AI industry, Together AI recently announced the completion of a $305 million Series B funding round, which has garnered widespread attention. The company's rise is closely related to its newly launched deep reasoning model, DeepSeek-R1. Contrary to initial concerns, many industry experts believe that advancements in deep reasoning have not diminished the demand for infrastructure; rather, they are continuously enhancing this demand. Image source note: Image generated by AI, image licensed from service provider Midjourney since its establishment in 2023.

Today, xAI, led by Elon Musk, officially launched its latest model Grok3, which Musk described as currently the 'smartest AI on Earth.' It will compete with ChatGPT, Gemini, DeepSeek, and Claude. During the launch event, xAI showcased many benchmark tests, revealing that Grok3 outperformed Gemini2Pro, Claude3.5Sonnet, GPT4o, and Dee in mathematical, scientific, and coding benchmarks.

With the advancement of artificial intelligence technology, multi-agent systems are increasingly capable of handling complex tasks across various fields. These systems are composed of multiple specialized agents that collaborate to leverage their individual strengths and achieve common goals. Such collaboration excels in areas like complex reasoning, programming, drug discovery, and safety assurance, as the structured interactions between agents not only enhance problem-solving efficiency but also allow for mutual correction, thereby improving each agent's output. Research indicates that this collaborative approach often outperforms in tasks that require rigorous reasoning or fact verification.

A recent study released by the University of Munich, the Munich Machine Learning Center, and Adobe Research shows that 12 top AI language models, including GPT-4o, Gemini1.5Pro, and Llama-3.3-70B, face significant performance degradation in long-text conceptual reasoning tasks. Despite these models supporting context processing of at least 128,000 tokens, their ability to establish deep logical connections still has fundamental limitations. The research team developed NOLIMA (No Text Matching).

Recently, OpenAI made significant updates after launching its flagship AI models o3 and o3-mini, altering the way o3-mini responds. Now, this model not only answers users' questions but also showcases its thought process, providing greater transparency to users. This change marks an important step for OpenAI in enhancing user experience, making the use of AI tools more human-centered. The updated o3-mini, especially in 'high reasoning' mode.

Recently, AI researchers from Stanford University and the University of Washington successfully trained an AI reasoning model named S1, with a training cost of less than $50 and very low cloud computing credits required. This research was published last Friday, indicating that S1 performs comparably to OpenAI's O1 model and DeepSeek's R1 model in mathematical and programming ability tests. The code and data for S1 have been made publicly available on GitHub for other researchers to use. The research team reported that they drew from a

Recently, OpenAI's CEO Sam Altman announced that the new reasoning AI model o3-mini will soon be officially launched for ChatGPT users, including both free and paid users. This new model is the successor to the o1 reasoning model and has been continuously optimized since its internal beta testing began last December, aiming to provide users with a more efficient, user-friendly, and cost-effective advanced reasoning AI service. The o3-mini model has made significant progress in fine-tuning for specific tasks, enabling it to handle various complex problems more effectively.

On January 24, Baichuan Intelligence made a significant announcement with the launch of China's first full-scene deep thinking model — Baichuan-M1-preview. This model excels across three dimensions: language, vision, and search reasoning, marking another major breakthrough in the AI field. Baichuan-M1-preview has demonstrated outstanding performance on multiple authoritative evaluation sets, particularly in mathematical reasoning and coding tasks, surpassing many well-known models, including o1-preview. It is worth mentioning that,

The Dark Side of the Moon Company proudly launches its brand new SOTA (state-of-the-art) model—k1.5 Multimodal Thinking Model, marking a significant breakthrough in multimodal reasoning and general reasoning fields. This model not only possesses outstanding multimodal processing capabilities but also demonstrates exceptional general reasoning abilities, effectively addressing various complex tasks. The standout feature of the k1.5 model is its multimodal reasoning capability. It can process information from different modalities including text, images, and sounds simultaneously, providing more comprehensive and accurate results.

Today, the Alibaba Cloud Tongyi team officially released the new mathematical reasoning process reward model Qwen2.5-Math-PRM. This model offers two sizes, 72B and 7B, with performance significantly outperforming similar open-source process reward models, especially excelling in identifying reasoning errors. The 7B version of Qwen2.5-Math-PRM astonishingly surpasses the widely popular GPT-4o, marking an important step in Alibaba Cloud's research and development of reasoning models.

The Alibaba Qwen team recently published a paper titled 'Lessons Learned from the Development of Process Reward Models in Mathematical Reasoning' and introduced two new models in the Qwen2.5-Math-PRM series, featuring 7B and 72B parameters respectively. These models break through the limitations of the existing PRM framework in mathematical reasoning, significantly improving the accuracy and generalization ability of reasoning models through innovative techniques. Mathematical reasoning has long been a major challenge for large language models (LLMs), especially regarding errors in intermediate reasoning steps.

Today, iFlytek officially launched its first deep reasoning large model, the Xunfei Spark Deep Reasoning Model X1, trained on a fully domestic computing power platform. This innovative achievement marks another significant advancement for iFlytek in the field of artificial intelligence. The Xunfei Spark Deep Reasoning Model X1 demonstrates characteristics similar to human 'slow thinking' during problem-solving, while achieving industry-leading results with relatively low computing power, with several key metrics ranking first in the country. Particularly in Chinese mathematical capabilities, this model performs exceptionally well, showcasing strong technical prowess.

At the smart office product upgrade launch conference held on January 7, iFLYTEK announced the upcoming launch of its latest 'iFLYTEK Spark Deep Reasoning Model X1' on January 15. This release is set to infuse more powerful artificial intelligence capabilities into iFLYTEK's smart office products, expected to further expand the application boundaries of smart office solutions. As a leading artificial intelligence company in China, iFLYTEK has been committed to deeply integrating AI technology with office environments. The newly released X1 model aims to enhance the deep reasoning capabilities of intelligent products, enabling them to handle complex tasks.

The research team from The Chinese University of Hong Kong (Shenzhen) and Shenzhen Big Data Institute has recently launched a medical large language model (LLM) named HuatuoGPT-o1. This model is specifically designed for complex reasoning in the medical field, aiming to enhance the reliability of medical diagnosis and decision-making. Unlike previous LLMs that focused on mathematical reasoning, HuatuoGPT-o1 specializes in the medical domain by simulating the rigorous thought processes of doctors in real-world practice, paving new avenues for the development of medical AI.

OpenAI has partnered with AI luminary and Stanford professor Andrew Ng to launch a groundbreaking free course titled "Reasoning with o1." This course, taught by Colin Jarvis, Head of Strategic Solutions Architecture at OpenAI, aims to help developers gain an in-depth understanding of the advanced reasoning capabilities of the o series models. Course link: https://www.deeplearning.ai/short-courses/reasoning-wit