Chinese startup DeepSeek has officially introduced a new language model called V4 Pro, continuing to actively increase its presence in the rapidly growing artificial intelligence market. The updated version with the V4-Pro-0813 index received significantly improved capabilities for autonomous AI agents. The new model is already available to users through a software interface, a proprietary mobile application, and the company's web interface. At the same time, the developers plan to change the pricing plan to use the software interface for the V4 Pro and V4 Flash models. The organization intends to increase the total cost of access and introduce differentiated prices depending on the time of day, dividing them into peak and off-peak load hours.

Interest in the new release has noticeably increased after the publication of the results of independent tests of the more affordable V4 Flash model. In some specialized tests, it was able to outperform the preliminary April version of the V4 Pro, which clearly demonstrates the extremely rapid development of the company's entire model line. The startup became one of the key participants in the Chinese market after the high-profile launch of the R1 model at the beginning of 2017. At the same time, competition in the industry continues to grow steadily, as technology giants such as Moonshot AI, Zhipu AI, MiniMax, Alibaba and ByteDance are actively developing their own advanced solutions.

In parallel with the release of new products, DeepSeek significantly increases spending on computing infrastructure and expands its own team. According to Reuters, in July, the company was preparing a new major investment round with a business valuation of about seventy-four billion dollars. Earlier, in June, the startup had already successfully raised about seven billion four hundred million dollars. The organization also expects to at least double the total number of its employees. In particular, the company intends to significantly expand the teams involved in the maintenance of data centers and the development of AI agents. In addition, the startup is actively hiring specialized engineers to create its own specialized processor for artificial intelligence. The successful implementation of this ambitious project in the future will allow DeepSeek to significantly reduce its technological dependence on equipment from the American Nvidia and Chinese Huawei.