DeepSeek ships V4 Pro and raises API prices up to 12x with peak-hour billing

Cheap AI just got a real price tag. DeepSeek released V4 Pro and announced API increases of up to 12x, with peak-hour billing like an electricity grid.

The new rate card

On August 13, 2026 DeepSeek shipped DeepSeek-V4-Pro-0813, the production version of its agent-focused flagship, into its app and API. The price news landed the same day. From August 16 at 16:00 UTC, V4 Pro output tokens go from $0.87 to $3.96 per million during peak hours, and cache-hit input jumps from $0.003625 to $0.044 per million, roughly 12x.

The structure changes too. DeepSeek is introducing peak and off-peak billing: peak hours are 01:00-04:00 and 06:00-10:00 UTC, and everything outside them runs at half the peak rate. For API customers, the bill becomes a scheduling problem.

The move resolves a week-old cliffhanger. On August 6, DeepSeek warned prices would rise “significantly” and gave no numbers. These are the numbers.

A price war running in both directions

The timing is what makes this notable. The same day, Cointelegraph reported OpenAI and Anthropic cutting prices on their models under pressure from Chinese rivals. The budget king is raising prices while the premium labs discount.

For two years, DeepSeek’s rates anchored the low end of the market and forced margin pressure onto Western labs. If the anchor is now moving up while incumbents move down, the spread that defined the cheap-AI era is collapsing from both sides. Teams that built products assuming permanently cheap DeepSeek inference have three days of notice and an off-peak window to work with.