DeepSeek, a Chinese startup focusing on artificial intelligence technology, has announced the official release of an updated V4 Pro model, aiming to help it regain its lead among fast-growing local competitors. Now the company is expanding its staff, increasing computing power and attracting investment.
Photo credit: Solen Feyissa / unsplash.com
Developers note that the DeepSeek V4-Pro-0813 model significantly expands the capabilities of managing artificial intelligence agents – which can be accessed through APIs, mobile applications and web interfaces. Previously, the company also announced an increase in API access prices for V4 Pro and V4 Flash: charging now depends on load during peak hours. The community will be paying close attention to the latest version, as the low-cost V4 Flash suddenly performs better than the pre-release V4 Pro in independent testing. This is unusual since the Pro is designed to be a more feature-rich product, but it shows that DeepSeek is rapidly improving its technical solutions between pre-release and official launch.
DeepSeek captured public attention in early 2025 with the release of Inference R1, which gained widespread adoption and raised questions about its ability to build powerful AI systems at a fraction of the cost of U.S. rivals. Its leadership in China has since been challenged by numerous competitors, including Moonshot AI, Zhipu AI, MiniMax, Alibaba and ByteDance.
The bigger challenge for DeepSeek is turning its leadership position into a sustainable business. In June, the company attracted US$7.4 billion in investment; in July it was reported that it was preparing to seek a new round of financing with a valuation of approximately US$71 billion. This is a significant change for a company that has long shunned outside capital. Competing in AI is becoming increasingly expensive, requiring investments in AI accelerators, data centers and human resources. DeepSeek plans to double headcount across all divisions, including its data center and AI agency divisions. It was previously reported that the company has increased its hiring of chip developers and hopes to create its own model release accelerator to reduce dependence on Nvidia and Huawei.
If you find an error, select it with your mouse and press CTRL+ENTER.