DeepSeek Raises V4 Prices — Monetizing the Share It Won on Cheap Tokens
DeepSeek will raise API prices for V4-Flash and V4-Pro and introduce peak and off-peak rates — increases Reuters put in a 50% to 1,100% range depending on model, token type, and hour — effective 16:00 UTC on August 16. Peak/off-peak plus a brutal cache-hit reset is how a price winner starts collecting.
Corrected October 14, 2026: Removed an unsupported characterization from the opening.
DeepSeek’s pitch was frontier-adjacent tokens at commodity prices. Peak/off-peak plus a brutal cache-hit reset is how a lab that won share on price starts monetizing that share.
DeepSeek posted an API pricing update for V4-Flash and V4-Pro, adding peak / off-peak rates (off-peak is half of peak). Peak hours: 01:00–04:00 and 06:00–10:00 UTC. New prices take effect 16:00 UTC, August 16, 2026 (DeepSeek’s X post and docs). Reuters said the new rates run ~50% to ~1,100% above prior list depending on model, cache hit/miss, and hour. This is a first-party GTM / pricing ship — not a new model.
New list (api-docs.deepseek.com, per 1M tokens)#
| V4-Flash cache hit | V4-Flash cache miss | V4-Flash output | V4-Pro cache hit | V4-Pro cache miss | V4-Pro output | |
|---|---|---|---|---|---|---|
| Off-peak | $0.007 | $0.22 | $0.66 | $0.022 | $0.66 | $1.98 |
| Peak | $0.014 | $0.44 | $1.32 | $0.044 | $1.32 | $3.96 |
Prior vendor list commonly cited before the change: Flash $0.0028 / $0.14 / $0.28 (hit / miss / out); Pro $0.003625 / $0.435 / $0.87. Cache-hit Pro is the steepest step (+507% / +1,114% off-peak / peak vs the old $0.003625). Concurrency caps on the new card: Flash 2,500, Pro 500. Reuters dated the company statement August 13 and in one wire converted the effective instant to August 17 local; the primary clock is DeepSeek’s 16:00 UTC, Aug 16.
Limits#
- Whether off-peak still undercuts everyone after FX is open.
- Whether OpenRouter mirrors the new vendor list is unsettled.
- Reuters range is secondary aggregation across model × cache × hour.