DeepSeek Launches V4 Pro as Chinese AI Startup Ramps Up Hiring, Computing and Fundraising

Chinese artificial intelligence startup DeepSeek on Thursday formally released its V4 Pro model, stepping up its efforts to regain momentum in China's competitive AI market as the company expands its workforce, computing capacity and access to capital.

DeepSeek said its V4-Pro-0813 model delivers major improvements in AI-agent capabilities and is available through its API, app and web platforms. The company also announced higher API prices for both V4 Pro and V4 Flash, alongside a new pricing structure that differentiates between peak and off-peak usage.

The pricing changes mark an important shift for a company whose rapid rise was initially built in part on the ability to offer powerful AI models at relatively low cost. As demand for advanced models increases, DeepSeek is now seeking to balance competitive pricing with the substantial computing expenses required to operate increasingly capable systems.

The V4 Pro launch will be closely watched after DeepSeek's less expensive V4 Flash model unexpectedly outperformed an April preview of V4 Pro in several independent tests.

That result was notable because the Pro version is intended to be the company's more capable model. The performance of Flash suggested DeepSeek had made substantial improvements to its underlying technology between the preview and the formal release of V4 Pro.

The latest model also arrives at a critical point for DeepSeek, which has faced growing competition from a rapidly expanding group of Chinese AI developers.

DeepSeek became one of China's most closely watched AI companies after its R1 reasoning model gained global attention in early 2025. The model triggered a broader debate over whether advanced AI systems could be developed with substantially lower costs and computing requirements than those associated with leading U.S. technology companies.

That early advantage has since come under pressure.

Chinese competitors, including Moonshot AI, Zhipu AI, MiniMax, Alibaba, and ByteDance, have released increasingly capable models, narrowing the technological and commercial gap with DeepSeek.

The competitive environment has also changed the challenge facing DeepSeek. Its initial breakthrough established the company as a major AI player, but maintaining that position requires substantially more resources as model development becomes more expensive and rivals release new systems at a rapid pace.

From AI Breakthrough to Capital-Intensive Business

DeepSeek is now preparing for a major expansion in both funding and infrastructure. Reuters reported in July that the company was planning a fundraising round at a valuation of about $74 billion, just weeks after raising approximately $7.4 billion in its first external financing round in June.

The fundraising represented a significant departure for DeepSeek, which had historically operated with limited reliance on outside capital.

The shift shows how developing frontier models requires access to large quantities of advanced computing hardware, data-center capacity and specialized engineering talent. Even companies that initially distinguish themselves through capital-efficient development must spend heavily to train, deploy and improve models at scale.

DeepSeek's planned expansion reflects those pressures. The company has said it intends to at least double its workforce across several departments, including teams focused on data centers and AI agents.

The emphasis on AI agents is notable as the industry is increasingly moving beyond chatbots that simply respond to prompts toward systems capable of planning tasks, using software tools and completing multi-step assignments with limited human intervention.

DeepSeek's decision to highlight agent capabilities in V4 Pro therefore puts the company directly into one of the fastest-developing areas of AI competition.

DeepSeek Explores Its Own AI Chips

The startup is also seeking greater control over the hardware underpinning its AI systems. Reuters reported in July that DeepSeek had stepped up private recruitment of chip-design engineers to develop its own AI processor.

Developing proprietary chips could eventually reduce the company's dependence on external suppliers, including Nvidia and Huawei, while giving DeepSeek greater control over the hardware needed to train and run its models.

The effort is part of a broader push across China's technology industry to reduce exposure to foreign semiconductor supply chains. Access to advanced AI chips has become a strategic issue as U.S. export controls restrict the availability of some high-end processors to Chinese companies.

Building competitive AI hardware, however, is considerably more difficult than designing an AI model. It requires semiconductor design expertise, software optimization, manufacturing partnerships, and access to advanced fabrication capacity.

Many believe that if DeepSeek succeeds, its hardware initiative could provide a long-term advantage by allowing the company to optimize computing infrastructure around its own models.

The V4 Pro launch therefore represents more than another model release.

DeepSeek is attempting to move from an AI startup that gained global attention through a technological breakthrough into a larger company capable of sustaining an expensive development and commercialization cycle. Its higher API prices, planned fundraising, workforce expansion, and investment in computing infrastructure all point in the same direction: the company is preparing for a much larger operating footprint.

The challenge is that DeepSeek no longer operates in the relatively open field that existed when R1 first attracted global attention. Chinese competitors are releasing models at a rapid pace, while U.S. companies continue to advance their own systems and expand the capabilities of AI agents.

That makes the performance and adoption of V4 Pro particularly important.