Back

DeepSeek Slashes API Cache-Hit Input Pricing by 90% Across Model Lineup

According to DeepSeek’s official website, cached input pricing was cut to one-tenth of launch levels, with a 75% limited-time DeepSeek-V4-Pro discount running through May 5, 2026.

Summary

DeepSeek officially announced that cached input pricing has been reduced to one-tenth of launch levels across its lineup. According to DeepSeek’s official website, DeepSeek-V4-Pro cached input pricing was cut to CNY 0.025 per million tokens, and the limited-time 75% discount will remain in effect until Beijing time 2026/05/05 23:59. In API billing, cache-hit pricing applies when previously processed input is reused, which can lower costs for repetitive or large-scale workloads.

Terms & Concepts
  • API (application programming interface): A set of rules that lets software systems connect and exchange data or services programmatically.
  • cache hit: A successful reuse of previously stored data, which can reduce processing time and input costs in repeated requests.
  • token: A basic unit of text processed by an AI model, often used to measure usage and pricing.