DeepSeek API pricing

Official rates in USD per million tokens. Peak hours cost double.

DeepSeek Flash

Item	Off-peak (50%)	Peak
Input, cache hit	$0.003	$0.006
Input, cache miss	$0.15	$0.30
Output	$0.60	$1.20

1M context, 384K max output, concurrency 2500. Supports vision input.

DeepSeek V4 Pro

Item	Off-peak (50%)	Peak
Input, cache hit	$0.022	$0.044
Input, cache miss	$0.66	$1.32
Output	$1.98	$3.96

1M context, 384K max output, concurrency 500. No vision input.

When peak pricing applies

Period	UTC	Beijing time	Rate
Peak	01:00–04:00, 06:00–10:00 (Mon–Fri)	09:00–12:00, 14:00–18:00	Standard
Off-peak	Everything else	Everything else (incl. weekends, holidays)	50%

Public holidays in China are excluded from peak. Converted to Beijing time, peak hours land exactly on the Chinese working day — that is the point.

Other billable APIs

API	Unit	Price
Search	per 1,000 requests	$0.75
Fetch	per 1,000 requests	$0.75

Cache hits are the biggest lever

A cache hit costs 1/50th of a cache miss. Put fixed content — system prompts, knowledge-base prefixes, long documents — at the front of the request and you stop paying for it on every call.