DeepSeek API Input Cache Price Slashed to One-Tenth of Original
The leading domestic large language model, DeepSeek, recently announced a significant price cut, reducing the input cache hit price across all API series to one-tenth of the original rate. This move marks a new phase in cost management for domestic AI, aiming to attract more developers and businesses by offering exceptional value for money.
Core Price Cuts Address Industry Pain Points
This price adjustment covers the entire V4-Pro and V4-Flash series. The input cache price for V4-Pro has been reduced to 0.1 RMB per million tokens, and with a limited-time promotion, the actual payment is only 0.025 RMB. Compared to overseas competitors, the input cache price is just 1/700 of GPT-5.5 Pro, demonstrating strong market competitiveness.
In addition to cache hit scenarios, prices for cache miss and output scenarios have also been reduced to one-quarter of the original price. This pricing strategy precisely targets high-frequency use cases such as RAG knowledge bases, intelligent customer service, and document analysis, potentially cutting enterprise operational costs by over 90%.

DeepSeek's ability to significantly reduce prices stems from its self-developed sparse attention architecture. This technology supports ultra-long context processing of up to 160K, improving efficiency in handling long texts while effectively lowering underlying computing power consumption and storage costs.
Related article
Google Launches Secure AI Tool to Challenge Ansopek in Code Face-Off
During the recent I/O Developer Conference, Google unveiled a significant cybersecurity initiative. The company invited a select group of experts to perform API testing on CodeMender, an AI agent designed for code security.Developed by Google DeepMin
How to write SEO titles for Google Japan in 2025?
SEO content writers face a tough spot. The industry's economics push you toward high output, and AI enables that volume. But AI-generated content that hasn't been thoughtfully humanized is becoming a growing risk—not just for the reader's experience
NDRC Deploys Embodied Intelligence Training Infrastructure for Large and Small Brain Models
At a recent press conference, Li Chao, Deputy Director of the Policy Research Office of the National Development and Reform Commission (NDRC), announced that the next phase will focus on advancing high-quality development in embodied intelligence, in
Related Special Topic Recommendations
Comments (0)
0/500
The leading domestic large language model, DeepSeek, recently announced a significant price cut, reducing the input cache hit price across all API series to one-tenth of the original rate. This move marks a new phase in cost management for domestic AI, aiming to attract more developers and businesses by offering exceptional value for money.
Core Price Cuts Address Industry Pain Points
This price adjustment covers the entire V4-Pro and V4-Flash series. The input cache price for V4-Pro has been reduced to 0.1 RMB per million tokens, and with a limited-time promotion, the actual payment is only 0.025 RMB. Compared to overseas competitors, the input cache price is just 1/700 of GPT-5.5 Pro, demonstrating strong market competitiveness.
In addition to cache hit scenarios, prices for cache miss and output scenarios have also been reduced to one-quarter of the original price. This pricing strategy precisely targets high-frequency use cases such as RAG knowledge bases, intelligent customer service, and document analysis, potentially cutting enterprise operational costs by over 90%.

DeepSeek's ability to significantly reduce prices stems from its self-developed sparse attention architecture. This technology supports ultra-long context processing of up to 160K, improving efficiency in handling long texts while effectively lowering underlying computing power consumption and storage costs.
Google Launches Secure AI Tool to Challenge Ansopek in Code Face-Off
During the recent I/O Developer Conference, Google unveiled a significant cybersecurity initiative. The company invited a select group of experts to perform API testing on CodeMender, an AI agent designed for code security.Developed by Google DeepMin
How to write SEO titles for Google Japan in 2025?
SEO content writers face a tough spot. The industry's economics push you toward high output, and AI enables that volume. But AI-generated content that hasn't been thoughtfully humanized is becoming a growing risk—not just for the reader's experience
NDRC Deploys Embodied Intelligence Training Infrastructure for Large and Small Brain Models
At a recent press conference, Li Chao, Deputy Director of the Policy Research Office of the National Development and Reform Commission (NDRC), announced that the next phase will focus on advancing high-quality development in embodied intelligence, in





Home






