DeepSeek adds peak and off-peak API prices; off-peak costs half

DeepSeek now charges two prices for its API, depending on the time of day. It announced the change on August 13, 2026, with the general release of DeepSeek-V4-Pro, and the new prices took effect at 16:00 UTC on August 16, 2026. Off-peak rates are half the peak rates.

When peak hours are

Peak pricing applies Monday to Friday, 01:00-04:00 and 06:00-10:00 UTC, except on Chinese public holidays. Every other hour, weekends and holidays included, is off-peak.

For US users, that means:

US time zone	Peak hours (local time)
Eastern, until Nov 1	9 p.m.-midnight and 2-6 a.m.
Eastern, from Nov 1	8-11 p.m. and 1-5 a.m.
Pacific, until Nov 1	6-9 p.m. and 11 p.m.-3 a.m.
Pacific, from Nov 1	5-8 p.m. and 10 p.m.-2 a.m.

The windows start on Sunday evening in US time and end early Friday. A normal US working day falls entirely in off-peak hours.

What it costs now

Per 1M tokens, from DeepSeek’s pricing page on October 18, 2026:

Model	Cache hit (off-peak / peak)	Input (off-peak / peak)	Output (off-peak / peak)
deepseek-v4-pro	$0.022 / $0.044	$0.66 / $1.32	$1.98 / $3.96
deepseek-flash	$0.003 / $0.006	$0.15 / $0.30	$0.60 / $1.20

Both models have a 1M-token context window. deepseek-v4-pro runs DeepSeek-V4-Pro-0813; deepseek-flash runs DeepSeek-V4.1-Flash-1017, updated on October 17 with no price change. The old deepseek-v4-flash names are retired and their requests are billed at the Flash price.

DeepSeek’s notices give the off-peak ratio but don’t state the prices that applied before August 16.

Coming next

DeepSeek’s Search and Fetch tools cost $0.75 per 1,000 successful requests. Calls made through DeepSeek Harness, Claude Code or Codex are free for now; billing for those starts on November 1, 2026 at 00:00 Beijing time.

What to do

If you run batch jobs on DeepSeek, schedule them outside the two peak windows: the same job costs half as much. When you compare DeepSeek with other APIs, use the price for the hours you actually run.