DeepSeek raises V4 AI model prices with peak-hour billing in mid-August

DeepSeek is sharply raising prices for its V4 AI models by introducing peak and off-peak billing, with some peak-hour charges more than quadrupling and, in some cases, rising far more depending on the token type. The change affects the V4-Pro and V4-Flash models and takes effect in mid-August, with peak periods set around Beijing business hours. The move narrows DeepSeek’s price advantage over larger AI competitors and signals higher demand or tighter capacity in its service, which matters for developers and businesses that rely on its API.