Overview
- DeepSeek publicly announced a plan to increase overall pricing for its API services and described the expected rise as substantial while saying exact rates will appear in a formal notice.
- The company advised customers to reasonably arrange their usage now to limit costs before the new price schedule is released.
- Current billing charges separate input tokens and output tokens and apply different rates for cache hits versus cache misses, a structure that DeepSeek highlighted in its advisory.
- IT之家 reported the baseline prices for two main models, listing DeepSeek‑V4‑Flash as low-cost for high concurrency and DeepSeek‑V4‑Pro as higher‑cost for complex reasoning, with the published example rates reflecting big gaps between cache hits and misses.
- Because the firm has not published final rates, businesses and developers face uncertainty about their bills and should watch for the formal notice while testing cache and usage changes that could blunt cost increases.