DeepSeek Slashes API Cache Pricing to 1/10th Across Entire Model Series

DeepSeek API Docs·medium signal

DeepSeek announced an immediate reduction in input cache hit pricing to one-tenth of the original price across its entire API series, including V4-Pro and V4-Flash. Combined with V4-Pro's $1.74/M input base price, the cache discount makes repeated-context workloads — agent loops, RAG pipelines, long conversation threads — dramatically cheaper. For builders running agentic systems with heavy context reuse, the effective per-token cost approaches negligible on DeepSeek infrastructure.