DeepSeek, the Hangzhou-based artificial intelligence company that disrupted the AI industry with remarkably low-cost language models, is implementing substantial price increases across its API offerings. Starting August 16 at 16:00 UTC, the company will raise prices for its V4-Flash and V4-Pro models by anywhere from 50% to more than 1,100%, depending on token type and time of day.

The restructuring introduces a new peak and off-peak pricing tier system. API usage during high-traffic hours will cost twice as much as usage during off-peak windows. The tiered approach mirrors demand-management strategies used by cloud infrastructure providers, where dynamic pricing helps allocate compute resources during periods of concentrated API traffic — a consideration particularly relevant for large language model inference, which remains computationally intensive even as training costs have declined.

Detailed Pricing Breakdown

For the V4-Flash model, off-peak cache-hit input tokens will cost $0.007 per million, up from $0.0028 — a 150% increase. Cache-miss input tokens will rise from $0.14 to $0.22 per million, representing approximately a 57% increase. Output tokens will increase from $0.28 to $0.66 per million during off-peak hours.

Peak rates will double all off-peak figures. Output tokens during high-traffic windows will run $1.32 per million, nearly five times the previous flat rate of $0.28 per million. The V4-Pro models will follow a similar tiered structure, with peak rates set at twice the off-peak level.

Even after the increases, DeepSeek's rates remain positioned at the lower end of the commercial LLM API market, where leading providers such as OpenAI and Anthropic charge significantly higher per-million-token rates for their frontier models. The adjustment narrows — but does not eliminate — the pricing gap that made DeepSeek a focal point for developers and enterprises seeking lower-cost inference.

Timing and Context

DeepSeek signaled the upcoming adjustment roughly ten days in advance. On or around August 6, the company indicated that price changes were forthcoming, citing heightened demand as the primary driver.

Founded in 2023 by Liang Wenfeng, DeepSeek operates under a funding structure that distinguishes it within the competitive AI landscape. The company is backed exclusively by High-Flyer, a Chinese quantitative hedge fund. DeepSeek gained widespread attention for delivering large language models with capabilities competitive with leading Western alternatives at a fraction of the cost, prompting discussion across the global AI sector about cost efficiency and accessibility. The price increase marks a notable inflection for a company whose ultra-low pricing strategy was central to that disruption, and developers building on DeepSeek's API will likely be watching whether further adjustments follow as demand continues to scale.