On September 9, DeepSeek Open Platform made an announcement stating that the V4.1 Flash model is slated for an official rollout around September 10, 2026 (Beijing Time). This new model boasts a cutting-edge architecture and offers native support for multimodal capabilities. According to test results, it outperforms the V4 Pro in several key areas, including overall performance, cost-efficiency, processing speed, and total turnaround time. In a parallel move, DeepSeek has decided to revise its pricing strategy for the Flash series. Starting at 12:00 PM on September 10, the cost per unit for input cache hits during off-peak periods will be slashed to 0.02 yuan, marking a significant 60% reduction.
