DeepSeek raised the prices of its API services for its latest V4 models. The new rates for V4-Pro and V4-Flash will come into effect on August 17, 2026, according to a company statement. The price increase will vary depending on the model, type of token, and time of use. Under the revised structure, some API rates will rise by 50%, while others will increase by as much as 1,100% compared with current prices.
DeepSeek will also introduce separate peak and off-peak pricing, giving customers different rates depending on when they use the services.
DeepSeek’s decision is notable since the company has built much of its reputation around offering powerful models at relatively low prices. Its competitive pricing helped to attract developers and businesses looking for alternatives to services from larger technology companies.
The new pricing system changes that equation, particularly for customers that rely heavily on the API to power applications, software products, or automated services. While the exact increase will depend on usage and model selection, the introduction of peak and off-peak rates means businesses may also have to rethink when they send requests to DeepSeek’s systems.
API pricing is an important consideration for developers since costs can quickly increase when an application handles large volumes of requests. Companies using language models for customer support, coding tools, content processing, or other software services typically calculate their expenses based on the number of tokens processed.
DeepSeek’s higher rates could therefore have a direct impact on businesses that have built products around its V4 models. The change also came as demand for advanced computing services increased. AI companies face substantial expenses related to computing infrastructure, data centres and the hardware needed to operate increasingly sophisticated models.
For DeepSeek, adjusting API prices could help the company manage those rising operating costs while maintaining access to its latest models.
Also Read: Alibaba's Qwen 3.8 Max Signals China's Next Big Leap in AI
The Chinese startup has been expanding beyond its reputation as a low-cost model provider. Earlier this month, DeepSeek had resumed a funding round targeting nearly $8 billion, with the company reportedly seeking a valuation of about 500 billion yuan ($74 billion). The company recently invested $20.8 million in Chinese robotics firm Unitree and agreed to collaborate on AI models for humanoid robots.
The new API pricing could consequently signal a broader shift as DeepSeek develops its commercial business. For developers, the immediate question will be whether the company's V4 models continue to offer enough performance and value despite the higher costs.