Cui Tianyi, a member of the DeepSeek team, stated in a post that V4.1 Flash outperforms V4 Pro in both performance and cost. The platform planned to route requests from V4 Pro directly to V4.1 Flash before the launch of V4.1 Pro, with only one day's notice and no migration buffer period provided, a move that sparked dissatisfaction among developers. Developers are concerned that a direct model switch could lead to business issues, while research teams have also pointed out that the deprecation of the old model will impact their research progress. Previously, Cui Tianyi also announced that DeepSeek would open around 150 positions for senior backend and server-side engineers to strengthen system engineering under the models, such as infrastructure construction for DSec. The core of this controversy lies in the decision-making power over model switches. In B2B operations, model stability often outweighs performance and price. Therefore, DeepSeek needs to improve its communication mechanisms and establish a secure migration engineering process to avoid eroding user trust.