DeepSeek warns of a significant API price increase without giving numbers or a date

DeepSeek told developers its API prices are going up significantly. It did not tell them by how much, or when.

What the notice says

On August 6, DeepSeek posted a notice that API prices will rise substantially across the board and urged users to plan ahead. There is no new rate card and no effective date. Details were promised in a separate announcement.

It is the second pricing move in under a month. Peak and off-peak pricing arrived in mid-July. As of the notice, V4 Flash costs $0.14 per million input tokens and $0.28 per million output tokens, the rates that made DeepSeek the default budget option for thousands of production apps.

Why the missing numbers are the problem

Price increases are ordinary. Announcing one without a figure or a date is what makes this notable for anyone with production traffic on the API. A migration is a scoped piece of work: it needs a target model, a cost delta to justify it, and a deadline to plan around. A warning that prices will go up significantly, eventually, gives a team none of those inputs while still forcing them to start the conversation.

Developer forums spent the day split between panic and shrugs, which is the predictable outcome when the only actionable content in an announcement is its tone.

The competitive backdrop

The timing sharpens it. A week earlier, OpenAI cut GPT-5.6 Luna prices by 80 percent. The cheap-model tier that DeepSeek effectively defined is now being contested by a lab moving in the opposite direction on price, at the moment DeepSeek is preparing customers for an increase.

That does not mean DeepSeek is wrong to raise prices. Inference at those rates has to be paid for somewhere, and two pricing changes in a month suggest a company actively reworking unit economics rather than drifting. But price was the whole argument for a large share of its production users, and the notice asks them to keep waiting without knowing what the argument will be next month.