Deepseek Launches V4 AI Model with 1.6 Trillion Parameters

Chinese AI startup Deepseek has released its new model, Deepseek V4, which features over 1.6 trillion parameters—more than twice the size of its predecessor V3.2. This launch follows an impactful debut in 2025, although it is expected to cause less disruption to the market this time. Both versions are open-source, emphasizing Deepseek's commitment to accessibility in AI technology.

The introduction of Deepseek V4 signifies a substantial shift in AI capabilities, particularly in autonomous functions like code generation. With these advancements, Deepseek may increase competition with established tech giants, highlighting the growing potential for national AI strategies to enhance local talent and reduce reliance on foreign technologies. This development signals a transformative moment for the AI landscape in China and the prospects for domestic innovation.