DeepSeek helped turn low-cost AI into one of China’s strongest competitive weapons. Its next move could show how long that pricing strategy can last.
The Hangzhou-based AI company said it plans to raise prices across its application programming interface services “in the near future,” with a “significant increase” expected.
DeepSeek has not disclosed the new rates, effective date, or affected models, leaving developers and businesses across China and the wider APAC region uncertain about how much more its services could cost.
The announcement arrived about a week after DeepSeek released V4-Flash-0731, a 284-billion-parameter lightweight model that quickly drew global interest for combining strong performance with unusually low inference prices.
How cheap is DeepSeek today?
DeepSeek’s official pricing page lists V4-Flash at $0.14 per million cache-miss input tokens and $0.28 per million output tokens. Cached input costs $0.0028 per million tokens. V4-Pro costs $0.435 per million cache-miss input tokens and $0.87 per million output tokens.
DeepSeek’s pricing has helped establish the company as a disruptive force in China’s crowded AI market. Artificial Analysis estimated that V4-Flash costs about $0.03 per task on average, compared with $3.15 for Anthropic’s Claude Fable 5, according to the South China Morning Post.
SCMP noted that Epoch AI also said DeepSeek’s pricing was an order of magnitude below Zhipu AI’s GLM-5.2 first-party API. The gap helped establish DeepSeek as a low-cost option for developers seeking capable open-weight models without paying the rates charged by larger US providers.
DeepSeek’s price move could reset China’s AI market
AI Weekly said that DeepSeek has not publicly explained why it is raising prices. The timing follows surging demand for V4-Flash-0731 and another pricing change in July that introduced peak and off-peak rates.
For Chinese AI companies, DeepSeek’s next price table could affect more than one vendor’s margins.
Its aggressive pricing reportedly pushed competitors including ByteDance and Tencent to lower their own rates. A substantial increase could give those companies room to raise prices or an opportunity to retain discounts and attract developers reconsidering DeepSeek.
The effects could also reach APAC businesses using Chinese AI infrastructure to control inference spending.
Teams that selected DeepSeek mainly for cost may need to model higher usage expenses, review alternative providers, or reconsider how heavily their applications depend on one vendor’s pricing.
Is DeepSeek’s low-cost edge disappearing?
Not yet. A large percentage increase from such a low base could still leave DeepSeek cheaper than many competing services.
The company’s mixture-of-experts architecture and training methods also help reduce the computation and memory needed to run its models, University of Hong Kong professor Zhenhui Jack Jiang told the South China Morning Post.
The larger question is whether DeepSeek can raise prices without weakening the advantage that helped distinguish it in China and abroad. The answer will depend on the size of the increase, which services it covers, and whether Chinese rivals follow or hold their prices down.
Until DeepSeek publishes those details, its low-cost edge remains intact. The planned increase nevertheless signals that China’s AI price war may be entering a more expensive phase.
Want to see what DeepSeek launched as its pricing strategy shifted? Read how V4-Flash entered public beta with a major AI agent upgrade.