DeepSeek API recharge service recovery, some model calls price adjustment

Recently, the DeepSeek API Open Platform announced that its API recharge service has officially resumed. Previously, due to tight server resources, DeepSeek suspended the recharge function of API services on February 6.

While restoring the recharge, DeepSeek also adjusted the prices of some models to call. Among them, the discount period of deepseek-chat model has ended, and the V3 model call price based on it is adjusted to 2 yuan per million input tokens and 8 yuan per million output tokens per million output. The deepseek-reasoner model based on the R1 model is 4 yuan per million input tokens and 16 yuan per million output tokens.

According to DeepSeek's official explanation, Token is the smallest unit used by the model to represent natural language text, which can be words, numbers, or punctuation marks, etc. DeepSeek will measure and bill based on the total number of tokens input and output of the model, and the deduction fee is equal to the token consumption multiplied by the unit price of the model. When the user's recharge balance and the gift balance exist at the same time, the system will give priority to deducting the gift balance.

The recovery of DeepSeek API recharge service and price adjustment will undoubtedly have a certain impact on users who use their services. Users need to pay attention to the new price standards and reasonably arrange recharge and use plans according to their own needs.