DeepSeek-V4-Pro Flagship Model's 25% Discount Period Extended to May 31

DeepSeek's official website recently updated its API documentation, showing that the 25% discount for its flagship large model DeepSeek-V4-Pro will be extended to 23:59 on May 31, 2026. Previously, the discount was set to end on May 6.

In addition, the model names deepseek-chat and deepseek-reasoner will be deprecated, corresponding to the non-thinking and thinking modes of deepseek-v4-flash respectively; the input cache hit price for all models has been reduced to 1/10 of the launch price, effective from 20:15 on April 26.

DeepSeek-V4-Pro uses a Mixture of Experts architecture with a total of 1.6 trillion parameters and supports million-level long context. Due to high-end computing power constraints, the current Pro version's service throughput is limited. It is expected that after the mass launch of Ascend 950 super nodes in the second half of the year, the price will be further reduced.