DeepSeek slashes prices on new V4-Flash model amid China AI cost battle

DeepSeek launched the open-weight V4-Flash model, offering enhanced agent functions while reducing token costs by 50%. The company also postponed a dynamic-pricing scheme that would have raised fees during peak usage. Although the new offering still lags behind Moonshot’s Kimi K3, the aggressive pricing signals a bid for customers in a rapidly competitive Chinese AI market. Beijing, a major backer, has warned against a race-to-the-bottom that is already driving down prices in other tech sectors.