DeepSeek lifts AI model fees by up to fourfold starting mid-August

DeepSeek will increase charges for its flagship V4 AI models on Aug. 16, applying a peak-hour price that is more than four times the current level. Users of the V4-Flash model will pay $1.32 for one million output tokens during busy periods, dropping to $0.66 in off-peak windows, up from $0.28 previously. The V4-Pro offering will cost $3.96 per million tokens at peak times and $1.98 when demand is lower, versus the former $0.87 rate.

The Hangzhou-based AI lab explained that the tiered pricing aims to distribute computing load more evenly. While the adjustment narrows the gap with major AI providers such as Anthropic PBC, DeepSeek’s fees still undercut many rivals. The move arrives amid broader debates over AI service costs and comes as DeepSeek prepares for a large fundraising round and a potential IPO later this year.