#What is the new pricing structure for DeepSeek's V4 series?
The upcoming V4 series from DeepSeek is set to launch in mid-July, introducing a tiered pricing model that increases token rates during peak hours. This adjustment aims to manage demand effectively and optimize the service's performance.
The peak periods are scheduled from 9:00 to 12:00 and again from 14:00 to 18:00 Beijing Time. During these windows, the costs for both input and output tokens will double for the two models offered: deepseek-v4-pro and deepseek-v4-flash. For instance, currently, V4-Flash is priced at approximately $0.14 per million input tokens and $0.28 per million output tokens. In peak times, these rates will escalate significantly.
For the V4-Pro model, peak output pricing is around 12 yuan per million tokens, equating to about $1.76. The peak output pricing for V4-Flash rises to about 4 yuan per million tokens, or approximately $0.59. To assist users in planning their expenses, DeepSeek will provide a 24-hour notice prior to any price changes.
#What do the V4 models offer?
The V4 series made its first public appearance as a preview on April 24, 2026. Both models are built on a Mixture of Experts (MoE) architecture, which enhances processing efficiency. V4-Pro features an impressive 1.6 trillion total parameters, while V4-Flash possesses 284 billion parameters. They both support expansive context windows of 1 million tokens and have been trained on a dataset that exceeds 32 trillion tokens. Both models are distributed under an MIT license, which promotes open access and collaboration within the developer community.
#Why is surge pricing significant for AI developers?
The peak pricing aligns with late evening and overnight hours in the United States, roughly from 9 PM to 6 AM Eastern Time. This timing effectively positions American developers to utilize DeepSeek's V4 models during their typical working hours at off-peak rates. It allows them to maintain cost efficiency while leveraging advanced AI capabilities in their projects. Understanding this pricing model can be crucial for developers looking to maximize productivity and budget for their AI developments.
In conclusion, DeepSeek's new pricing structure and the specifications of the V4 series present significant considerations for any developers engaged in AI. The strategic timing of off-peak rates provides an advantageous opportunity for users to benefit from advanced processing capabilities without incurring peak costs.