The Chinese company DeepSeek officially launched its new artificial intelligence model, V4 Pro, on Thursday, the 13th. The company seeks to consolidate its position against local competitors while increasing its hiring, computational capacity, and fundraising efforts.
This new model, named V4-Pro-0813, features enhanced functionalities designed for AI agents and can be accessed via API, application, or web. Furthermore, DeepSeek released new pricing tables for the APIs of the V4 Pro and V4 Flash models.
The announcement gained more visibility after the V4 Flash, a more economical version from the company, demonstrated superior performance compared to the preview of V4 Pro, presented in April, in certain third-party tests. This result generated interest, given that Pro was conceived to be DeepSeek's most sophisticated model.
According to Reuters, this performance suggests that the company may have significantly accelerated its technological progress between the preliminary phase and the official launch. Currently, DeepSeek faces more intense competition within China, with companies such as Moonshot AI, Zhipu AI, MiniMax, Alibaba, and ByteDance also launching new models.
DeepSeek gained global recognition at the beginning of 2025, driven by the success of the R1 model, which fueled discussions about the feasibility of creating advanced AI systems at costs lower than those practiced by North American corporations.
Need for Funding and Infrastructure
However, sustaining this prominence requires considerable capital and infrastructure. As reported by Reuters, the company was planning a new funding round with an estimated valuation of around $74 billion (equivalent to approximately 384 billion reais), after having raised about $7.4 billion (approximately 38.4 billion reais) in its first external round in June.
This fundraising signals a strategic shift for an organization that previously avoided seeking external capital. It also illustrates the growing cost inherent in the AI race, which includes acquiring chips, maintaining data centers, and hiring specialists.
DeepSeek aims to expand its workforce by at least double in various areas, including teams dedicated to AI agents and data centers. Reuters also reported in July that the company increased private hiring of engineers specializing in chip design, aiming to develop its own artificial intelligence chip and reduce dependence on suppliers like Nvidia and Huawei.
In summary, with V4 Pro, DeepSeek does not just aim to maintain the notoriety achieved with R1; the company is strengthening its structure, team, and investment capacity to face an increasingly fierce competition in the Chinese artificial intelligence market.