Underwhelming or underrated? DeepSeek V4 shows “impressive” gains

DeepSeek's V4 model has underperformed compared to its local and American competitors, ranking second among open-source models, behind Moonshot AI's Kimi K2.6. This setback poses a significant challenge for DeepSeek, which aimed to replicate the success of its previous model, R1. Following the remarkable success of R1, DeepSeek is striving to enhance its market performance. However, current challenges indicate difficulties in meeting rising market expectations. According to a report by Artificial Analysis, despite improvements, V4 Pro has not been sufficient to surpass the competition.