DeepSeek, the leading domestic large model, recently sent an upgrade reminder to users, announcing that its new generation DeepSeek V4 general version is scheduled to be launched in mid-July. Notably, the new version will introduce a refined peak and off-peak pricing mechanism for the first time, along with multiple functional optimizations and performance improvements.

API Call Fees Double During Peak Hours

According to the official email information, the peak hours are officially set as 9:00 to 12:00 and 14:00 to 18:00 Beijing Time daily. During these periods, the cost of API calls for developers and enterprises will be doubled compared to regular times.

Although the doubling of prices during peak hours may affect the gross profit margin of AI applications with high-frequency access, the base rate has already been extremely low due to a previous significant and permanent price reduction. Therefore, even after doubling, the overall calling cost of DeepSeek V4 is still much lower than that of overseas leading models.

Global Enterprises Enter an Era of Cost Management

Industry analysis points out that this adjustment does not mean DeepSeek has abandoned its popular low-price strategy, but rather marks the beginning of its refined operation of computing power resources. Currently, several Chinese large models, with their extremely high cost-effectiveness, are showing strong market competitiveness in the international market.

As the billing models of major overseas AI tools shift towards token usage, budget-limited overseas tech companies are accelerating their migration to low-cost open-source models. This tiered, on-demand multi-model routing strategy is driving global developers into a more rational phase of AI infrastructure planning.