Search AI Products and News

Recently, the domestic large model DeepSeek has gone viral across the internet, becoming the focus of global tech markets due to its technological advantages of 'low cost and high performance.' Founder Liang Wenfeng mentioned that the team is primarily composed of graduates and doctoral students from domestic universities. The impressive innovative achievements indicate that today's China is becoming fertile ground for top talent and a source of original innovation. According to Qichacha data, as of February 6, there have been a total of 16,400 patent applications related to large models. In terms of application years, these patents have been predominantly filed in the last two years.

DeepSeek has officially released and open-sourced its latest large language model R1, which has shown outstanding performance and is considered comparable to OpenAI's official version o1. This initiative not only marks another significant breakthrough in domestic AI technology but also provides new options for AI developers worldwide. DeepSeek R1 has extensively applied reinforcement learning techniques in the post-training phase, significantly enhancing the model's reasoning capabilities even with minimal labeled data, in areas such as mathematics, coding, and natural language reasoning.

Recently, the domestic large model DeepSeek V3 has drawn industry attention due to its outstanding performance in the AI arena. As the only open-source model to break into the top ten, it not only surpassed o1-mini but even exceeded Claude 3.5 Sonnet in various fields, including programming and mathematics. To verify its actual capabilities, a series of comparative tests were conducted. In the basic comprehension ability test, both models exhibited different characteristics. Faced with the Chinese riddle 'Xiaoming's mother has three children', DeepSeek V3 performed...

The Tencent Hunyuan Large Model excelled in the August rankings of the SuperCLUE-V evaluation benchmark for Chinese multimodal large models, securing the top position among domestic large models and placing in the 'Exemplary Leaders' quadrant. Multimodal understanding requires the model to accurately identify image elements, comprehend their relationships, and generate natural language descriptions, testing the model's precision in image recognition and its understanding of the complex real world.

At the World Artificial Intelligence Conference in Shanghai, Alibaba Cloud CTO Zhou Jingren announced the latest developments of the Tongyi Large Model and the Alibaba Cloud Baoyan Platform. The download volume of the Tongyi Qianwen open-source model has doubled in two months, exceeding 20 million times. At the same time, the number of clients using Alibaba Cloud Baoyan services has increased from 90,000 to 230,000, achieving a growth of over 150%.Zhou Jingren emphasized Alibaba Cloud's commitme

In 2023, domestic large models have risen, facing challenges in computing power, characterized by 'external sanctions and internal development'. The ban has pushed domestic GPU chips into the spotlight, with Huawei Ascend and Cambricon becoming the focus. The computing power issue remains severe, and public cloud providers are addressing computing demands by 'centralizing computing power and substituting cloud for cards'. Data has become a key factor in the quality of large models, and high-quality language data may be exhausted, with a scarcity of Chinese datasets. In the future, companies may need to invest in finding quality data, with platforms like Zhihu and Douban potentially seizing development opportunities.

Zhipu AI is reported to be seeking a new round of financing with a valuation of 20 billion RMB, making it one of the highest-valued companies in the domestic large model sector. It has launched the third-generation base model ChatGLM3, which competes with GPT-4V, showcasing significantly upgraded performance and impressive rankings. Investors include major companies such as Alibaba, Meituan, Tencent, and Xiaomi, as it enters the later stages of Series B financing, aiming to raise over 2.5 billion RMB. ChatGLM3 enhances performance and reasoning capabilities, introducing a new code interpreter feature that has garnered widespread attention. Zhipu AI demonstrates its high valuation and continuous financing success.

This article selects recent true questions from the Teacher Qualification Exam to test the answering abilities of three domestic large models. iFLYTEK Spark, Wenxin Yiyan, and Tongyi Qianwen performed well overall in subject knowledge and logical reasoning. However, there is still room for improvement in regulations and comprehensive quality. Wenxin Yiyan's responses were relatively standardized, demonstrating certain answering skills. Large models may assist in preparing for the teacher qualification exam, but manual review is still necessary.

{Point 1: iFlytek Spark has launched features such as image generation, image understanding, virtual human video generation, and code generation. Point 2: iFlytek Spark supports plugin functionalities, including document Q&A, PPT generation, and resume creation. Point 3: iFlytek Spark offers powerful code generation, explanation, and error correction capabilities to meet developer needs.}