Deepseek, an artificial intelligence company based in China, has announced an upgrade of its AI chatbot, saying that it can now offer a global logic, mathematics and improved programming with a reduced hallucination rate.
According to Deepseek, the improved model – Deepseek -R1-0528 – has “considerably improved its reasoning depth and inference capacities”. The startup said that the overall performance of the model “now tackles those of the main models, such as O3 and Gemini 2.5 Pro”.
The beginnings of Deepseek on his chatbot R1 in January sent shock waves in the AI industry and still established China as AI force. The company’s first AI model had a training cost of $ 6 million and performance similar to the main models of AI trained in much larger capital.
According to business of Business of Apps, Deepseek has been downloaded 75 million times since its launch and had 38 million monthly active users (MAU) in April. In a recent antitrust trial, Google estimated that Gemini reached 350 million active users in March, while the OpenAi chatgpt claimed 600 million active users during the same month.
In relation: Deepseek in China is launching the new Open-Source AI after R1 took Openai
https://www.youtube.com/watch?v=fykm3ka_lzu
The Chinese-American race AI heats up
The US government plans to restrict the sale of advanced flea design software to China. According to a Bloomberg report, this decision aims to limit China’s ability to advance its manufacturing capacities of interior semiconductors.
Semiconductors are essential for a wide range of technologies, including AI, where they serve as a material skeleton for complex training and management.
The new models of AI China, such as T1 and Qwen3 of Tencent and Alibaba, also emerged in the first months of 2025, stimulating the AI race.
Review: Ai Eye: 9 curious things about Deepseek R1
