Search
On April 26th, DeepSeek, a prominent domestic large-scale model provider, announced a significant price reduction for input caching hits across its entire API suite. The new pricing structure sets the cost at merely one-tenth of the initial launch price. Specifically, the Pro model is currently enjoying a time-limited discount, with an extra 75% reduction valid until May 5th. This brings the cached input price for the V4-Pro model down to an astonishingly low 0.025 yuan per million Tokens, and for the V4-Flash model to just 0.02 yuan per million Tokens, establishing a new benchmark for global large-scale model pricing. This price adjustment encompasses the entire range of DeepSeek-V4-Pro and V4-Flash models, with the most substantial reductions targeting input caching hit scenarios.
-
C114 Communication Network -
Communication Home
