Skip to content

Pricing & Cost

DeepSeek V4 Pro Pricing Shifts to Peak and Off-Peak

DeepSeek V4 Pro pricing stays cheap until 16.00 UTC on August 16, then splits into peak and off-peak. Off-peak is half of peak.

The Vibe Father 5 min read
DeepSeek whale symbol and wordmark on black
DeepSeek product wordmark. Editorial reference TheVibeFather media library Editorial reference
Share Post to X LinkedIn

DeepSeek V4 Pro pricing is about to stop being a single line. Until 16.00 UTC on August 16, 2026, the public sheet for deepseek-v4-pro is $0.003625 cached input, $0.435 cache miss, and $0.87 output per million tokens. After that moment, DeepSeek splits the day into peak and off-peak. Off-peak is half of peak.

Peak hours are 01.00 to 04.00 UTC and 06.00 to 10.00 UTC. Every other hour is off-peak. If you run agents on a US afternoon or a China evening, check which bucket you actually occupy before you celebrate the “half price” line.

The new Pro sheet

CellCurrentOff-peak after August 16Peak after August 16
Cache hit$0.003625$0.022$0.044
Cache miss$0.435$0.66$1.32
Output$0.87$1.98$3.96

Read that twice. Cache-hit and output both rise, even at off-peak, compared with today’s sheet. The GA model is more expensive to serve. DeepSeek is using time-of-day pricing to push batch work into quieter hours. Flash has the same structure at lower levels, $0.22 / $0.66 off-peak miss and output, $0.44 / $1.32 at peak.

How a coding team should react

  1. Move eval sweeps, backfills, and doc rebuilds into off-peak.
  2. Keep interactive coding on whatever hour the humans work, and budget peak.
  3. Turn cache keys on. The cache-hit line is still the bargain, even after the hike.
  4. Do not compare today’s $0.435 / $0.87 screenshot to Opus 5 next month without the new sheet.
  5. Watch the 500-concurrency Pro cap if you try to hide from peak by blasting off-peak.

Even after the hike, peak Pro output at $3.96 is still well under Grok 4.6 at $6 and Opus 5 at $25. The bargain is narrower, not gone. Pair this note with token-cost discipline and the live scorecard.

A simple planning rule is to price interactive work at peak and batch work at off-peak, then compare that blended number with Grok 4.6 and Flash. If the blended Pro bill still beats the next model you would actually use, keep Pro. If it does not, the GA quality bump is not free enough to ignore the calendar.

Bottom line

DeepSeek V4 Pro pricing on August 16 is a scheduled raise dressed as a peak and off-peak split. Model the new numbers now. If your agents can wait a few hours, off-peak still buys a lot of work. If they cannot, stop quoting April prices.

Sources

Reader check

Was this article helpful?

One click helps us decide what to research next.

The app behind this research

TheVibeFather is the multi-CLI AI coding harness

You just read field notes from the same team that ships TheVibeFather — the multi-CLI AI coding harness that runs Claude Code, Codex, OpenCode and more with shared memory and a verify gate. Bring your own keys.

Keep reading