DeepSeek Raises API Prices by Up to 1,100% Starting August 16 with New Peak/Off-Peak Billing

My take: For any company or team that has integrated DeepSeek's API into their workflow, this change has direct impact: starting August 16 at 16:00 UTC, V4-Flash output pricing goes from $0.28 to $1.32 per million tokens at peak hours (up 371%). V4-Pro rises from $0.87 to $3.96 per million at peak (up 355%). The maximum increase across tiers and token types reaches 1,100%.

The new billing model introduces peak and off-peak rates: low-demand hours cost half the peak rate (peak windows are 01:00-04:00 and 06:00-10:00 UTC). That means there is real room to absorb part of the increase by shifting heavier workloads to off-peak windows.

The most important signal from this story is not the price itself: it is that a single AI provider's pricing policy can shift dramatically from one month to the next. DeepSeek was the catalyst for the AI price war; now, under the pressure of demand, it is reversing course. Any AI architecture built on a single low-cost vendor, with no contingency plan, is exposed to exactly this kind of decision.

Does your team have a vendor diversification strategy for AI, or does it still depend entirely on one provider?