DeepSeek unveils V4-Pro and V4-Flash AI models in direct challenge to US tech dominance
New preview versions position the Hangzhou-based firm as a formidable competitor to Google and OpenAI, though data protection concerns persist
Chinese artificial intelligence startup DeepSeek has released preview versions of its latest generative models, DeepSeek-V4-Pro and DeepSeek-V4-Flash. The announcement marks a significant escalation in the global competition for technological supremacy, with the firm explicitly positioning its new systems as direct competitors to United States giants such as OpenAI and Google.
The DeepSeek-V4-Pro model asserts dominance in specific technical domains, claiming to outperform all rival open-source models in mathematics and coding. In terms of general world knowledge, the firm states the model trails only Google's Gemini 3.1-Pro. This performance claim is designed to counter the narrative that Western technology remains unassailable in critical computational fields.
Complementing the Pro variant, the DeepSeek-V4-Flash model is described as offering similar reasoning capabilities but with a focus on operational efficiency. The startup highlights that this version provides faster response times and more cost-effective pricing, aiming to lower the barrier to entry for developers seeking high-performance tools without the associated financial overhead.
Both new iterations maintain the open-source architecture established by the company's previous releases. This approach allows developers to utilise and modify the models freely, a strategy that has previously disrupted the Silicon Valley market. The release of these tools occurs exactly one year after the launch of DeepSeek-R1, which initially stunned the tech sector with capabilities broadly comparable to ChatGPT and Gemini.
The rapid iteration of these models recalls the controversy surrounding the January release of DeepSeek-R1. At that time, venture capitalist Marc Andreessen described the launch as "AI's Sputnik moment", noting the company's claim to have achieved high performance with computing costs under $6m. This assertion was a fraction of the multibillion-dollar budgets typical in the United States, though some analysts have since challenged the accuracy of these cost figures.
The geopolitical implications of this technological advancement remain significant, particularly regarding data sovereignty and national security. The previous model faced immediate backlash in several nations, leading to bans or restrictions in Australia, the United States, Taiwan, South Korea, Denmark, and Italy. These measures were implemented due to concerns over data protection and potential Chinese government censorship, issues that continue to frame the discourse surrounding the new V4 releases.
