DeepSeek Quadruples V4 Token Prices With Peak-Hour Plan

DeepSeek is rolling out peak-hour pricing for its flagship V4 models, a change that will roughly quadruple current rates for users who hit the busiest windows.

The new pricing takes effect Sunday, Aug. 16, Bloomberg reported Thursday. Under the plan, users of the DeepSeek-V4-Flash model will pay $1.32 per million output tokens during peak hours, with off-peak pricing set at half that. The current price is 28 cents per million tokens.

For the higher-tier V4-Pro model, peak-hour pricing will be $3.96 per million tokens, also half that off-peak, up from the current 87 cents per million tokens.

DeepSeek's new prices still undercut several competitors. Moonshot's Kimi K3 charges $15 per million output tokens. Anthropic's Fable 5 runs $50, according to the Bloomberg report.

In a Thursday changelog update, DeepSeek said the adjustment is meant "to allocate resources more reasonably." The company said it would use peak and off-peak pricing "with off-peak prices set at half of the peak-hour prices, encouraging users to schedule their tasks based on actual usage."

The pricing move comes as DeepSeek pursues a second funding round, reportedly aiming to raise close to $8 billion at a roughly $74 billion valuation. The company first drew wide attention in January when it released an AI model that delivered performance comparable to American rivals like OpenAI and Meta while using significantly fewer Nvidia chips.

DeepSeek has been expanding its product scope. On Wednesday, Bloomberg reported the company is building a new team to compete with Anthropic's Claude Code in the market for AI agents that automate business tasks. In April, DeepSeek offered developers a 75% discount on its V4-Pro model through May 5, a promotional push that now reads as a prelude to the new tiered pricing.

DeepSeek's Alpha Score sits at 77 out of 100, a "Strong" rating from AlphaScala's proprietary model.