The R1-0528 model has been introduced with improved reasoning capabilities, and DeepSeek asserts that the performance gap with the latest models from OpenAI and Google is being significantly reduced.
DeepSeek, a China-based artificial intelligence company, has announced an upgrade to its AI chatbot, highlighting improvements in logic, mathematics, and programming, along with a lower hallucination rate.
DeepSeek stated that the updated model, DeepSeek-R1-0528, has made notable strides in reasoning and inference capabilities. The company noted that its performance now nears the level of top-tier models like O3 and Gemini 2.5 Pro.
The January launch of DeepSeek’s R1 chatbot shook the AI sector and further solidified China’s presence in the artificial intelligence arena. Despite a relatively modest training cost of $6 million, the company’s inaugural model demonstrated performance comparable to top AI systems developed with substantially greater financial resources.
Data compiled by Business of Apps shows that DeepSeek has seen 75 million downloads since its debut, with 38 million monthly active users (MAU) recorded in April. Meanwhile, Google, in a recent antitrust filing, estimated Gemini’s user base at 350 million in March. During the same period, OpenAI reported 600 million active users for ChatGPT.
U.S.-China AI Rivalry Intensifies
The United States government is formulating plans to impose restrictions on exporting sophisticated chip design software to China. As reported by Bloomberg, the initiative aims to curb China’s progress in developing its own semiconductor production infrastructure.
Semiconductors play a vital role across numerous technologies, particularly in artificial intelligence, where they provide the foundational hardware required to train and operate intricate models.
Several Chinese companies, including Tencent with T1 and Alibaba with Qwen3, introduced new AI models in early 2025, further accelerating the competitive momentum in the artificial intelligence landscape.