DeepSeek became an international reference by showing that a Chinese laboratory could offer competitive models at a fraction of its rivals' prices. That promise has entered a different phase.
The company introduced V4 Pro and announced new API prices from 17 August. Reuters calculates increases ranging from roughly 50% to 1,100%, depending on model, time of use and whether input is cached.
The largest percentage does not describe every request. The schedule distinguishes V4 Flash and V4 Pro, input and output, cached data and peak or off-peak periods. The actual bill depends on how an application routes traffic.
That turns pricing into a design decision: overnight processes can move to cheaper hours, Pro can be reserved for complex work and Flash can handle classification or extraction. DeepSeek is selling not only intelligence but also a utilization curve.
AdvertisementARQUITHEAArchitecture for seeing more clearlyIdeas, buildings and tools for understanding the city through real questions.Follow @arquithea_ ↗For studios and creative teams, the lesson is not to abandon DeepSeek. It is to measure cost per accepted result, human review time, data terms and the effort required to switch providers before a promotional price becomes the architecture of the entire workflow.
Can DeepSeek remain the budget alternative while financing itself like a frontier AI company?
It can if Flash remains useful for high-volume work and the relationship between capability and cost stays clear. Customers should still preserve alternatives: a cheap API stops being cheap when changing provider requires rebuilding the whole system.
