DeepSeek Raises API Prices Fourfold With New V4 Pro Model Launch
DeepSeek has announced the upcoming launch of its DeepSeek V4 Pro model alongside a significant fourfold increase in its API pricing. The company is implementing a new peak and off-peak pricing structure scheduled to begin on August 16.
Artificial intelligence firm DeepSeek has officially announced the upcoming release of its new DeepSeek V4 Pro model, a development that coincides with a major restructuring of the company's financial framework for developers. According to the announcement, DeepSeek is raising its application programming interface pricing fourfold as part of a newly introduced peak and off-peak pricing structure.
The updated cost schedule, which is set to take effect on August 16, marks a sharp departure from the company's previous rates. Under the newly unveiled pricing tiers, the DeepSeek V4 Pro model will cost $3.96 for every 1 million output tokens during peak hours. However, developers utilizing the service during designated off-peak hours will benefit from a reduced rate, with the V4 Pro pricing dropping to $1.98 for 1 million output tokens.
In addition to the flagship V4 Pro model, DeepSeek has outlined pricing details for the DeepSeek V4 Flash model. The Flash model will cost $1.32 for 1 million output tokens during peak hours, up from its previous rate of $0.28. During off-peak hours, the DeepSeek V4 Flash model will be available at a cost of $0.66 for 1 million output tokens.
For context within the broader competitive market landscape, other industry options command varying rates for output tokens. Competitor Moonshot offers its Kimi K3 model at a cost of $15 per 1 million output tokens. Meanwhile, offerings from OpenAI feature distinct pricing structures, with OpenAI's GPT-5.6 Sol priced at $30 for 1 million output tokens, while OpenAI's GPT-5.6 Luna is priced at $1.20 for the same volume.
The transition to higher pricing follows a period of temporary promotional rates. DeepSeek's current API pricing was previously scheduled to conclude as a promotion ending on May 31. Although DeepSeek initially announced that it would make the discounted prices permanent, the company has ultimately decided to proceed with the substantial price hikes.
Explaining the rationale behind the upcoming adjustments, DeepSeek stated that the new peak and off-peak pricing structure is designed to allocate resources more reasonably. The implementation of these changes on August 16 will permanently alter how developers access and utilize DeepSeek's expanding suite of AI models.
