DeepSeek is preparing a sharp increase in API pricing for its latest V4 models, according to Engadget, moving from its unusually low-cost positioning toward a peak and off-peak structure starting August 16. Engadget reports that, with the announcement of DeepSeek V4 Pro, the company is raising API pricing roughly fourfold. The company said the new pricing structure is meant to “allocate resources more reasonably,” according to the report. The headline change is for DeepSeek V4 Pro output tokens. Starting August 16, Engadget reports that V4 Pro will cost $3.96 per 1 million output tokens during peak hours, up from $0.87. During off-peak hours, the same output will cost $1.98, or half the peak-hour rate. DeepSeek V4 Flash is also getting more expensive. Engadget reports that Flash will cost $1.32 per 1 million output tokens at peak hours, up from $0.28, and $0.66 during off-peak periods. The move matters because DeepSeek built much of its profile around lower-cost AI services compared with larger Western rivals. Engadget notes that the current API pricing had originally been framed as a promotion ending May 31, and that DeepSeek later said the discounted prices would become permanent before ultimately moving ahead with higher pricing. Even after the increase, Engadget reports that DeepSeek’s V4 pricing remains below some rival models. The report cites Moonshot’s Kimi K3 at $15 per 1 million output tokens and OpenAI’s GPT-5.6 Sol at $30. But not every comparison favors DeepSeek: Engadget says OpenAI’s GPT-5.6 Luna costs $1.20 per 1 million output tokens, below DeepSeek V4 Flash’s peak-hour price. The available evidence is a single reputable report, so the pricing specifics should be treated as reported rather than independently confirmed by this cluster. The concrete operational change, if the schedule holds, is that developers using DeepSeek’s V4 APIs will need to account for both higher rates and time-of-day pricing beginning August 16. Who benefits: DeepSeek may benefit from higher API revenue and from using time-based pricing to manage demand, based on its stated resource-allocation rationale. Users with flexible workloads may benefit relative to peak-hour users if they can run jobs during off-peak periods. Who's exposed: Customers with applications built around DeepSeek’s current low V4 API prices are exposed to higher inference costs. Peak-hour workloads are the most directly affected under the reported schedule.