DeepSeek signals API price hikes

DeepSeek told developers on August 6 that it plans to raise prices across its API services, according to a notice posted on the company’s pricing page. The Hangzhou-based company said the increase is expected to be “significant,” but it did not publish a new rate card or an effective date. The notice appeared less than a week after DeepSeek’s July 31 public-beta release of DeepSeek-V4-Flash-0731 and after another pricing adjustment in July, according to company documentation and outside reports. The move adds fresh uncertainty for customers that have used DeepSeek as a low-cost model supplier. ### What exactly did DeepSeek say? DeepSeek’s pricing page said on August 6: it plans to raise “the overall pricing for DeepSeek API services in the near future, with a significant increase expected.” The company added that the “specific pricing plan will be subject to official notice” and told users to “plan your usage accordingly.” (api-docs.deepseek.com) The same page said product prices may change and that DeepSeek reserves the right to adjust them. That language sits alongside the current posted rates for its main API models. ### What prices are posted right now? DeepSeek’s current public rate card lists DeepSeek-V4-Flash at $0.0028 per 1 million input tokens on cache hit, $0.14 per 1 million input tokens on cache miss, and $0.28 per 1 million output tokens. (api-docs.deepseek.com) DeepSeek-V4-Pro is listed at $0.003625 per 1 million input tokens on cache hit, $0.435 per 1 million input tokens on cache miss, and $0.87 per 1 million output tokens. The pricing page also says the rates are quoted per 1 million tokens and that billing is based on total input and output token use. DeepSeek separately lists concurrency limits of 2,500 for V4-Flash and 500 for V4-Pro. ### Why are users focusing on the timing? Dataconomy reported on August 6 that the warning is DeepSeek’s second pricing change in less than a month, after the company introduced peak and off-peak rates in mid-July. (api-docs.deepseek.com) TechNode and Bloomberg also reported that DeepSeek had warned users of a substantial increase without disclosing the final schedule or amount. DeepSeek’s own changelog shows that July 31 brought the public-beta release of DeepSeek-V4-Flash-0731, while the company’s pricing page says support for DeepSeek-V4-Pro in the Responses API is due in early August 2026. Those product changes landed close to the new pricing warning. ### How does this fit with DeepSeek’s position in the market? (dataconomy.com) Bloomberg said DeepSeek’s planned increase is unusual for a company whose low-cost models had put pressure on U.S. and Chinese rivals. Dataconomy said DeepSeek’s earlier low pricing helped push Chinese competitors including ByteDance and Tencent to cut their own prices after DeepSeek introduced a permanent V4 discount in May. (api-docs.deepseek.com) DeepSeek’s API documentation says its service is compatible with OpenAI- and Anthropic-style interfaces, a setup that has made it easier for developers to route workloads through its models. That compatibility means price changes can affect customers that have treated model choice as a switching decision rather than a full platform migration. (bloomberg.com) ### What is still missing from the announcement? DeepSeek did not say on August 6 whether the increase will apply evenly across V4-Flash, V4-Pro, cache-hit pricing, cache-miss pricing, or output pricing. The company also did not publish an effective date, a percentage increase, or any transition period for existing users. (api-docs.deepseek.com) The next formal milestone is DeepSeek’s promised “official notice” with the specific pricing plan. Until that notice appears on the company’s pricing page or developer documentation, customers only have the current posted rates and the August 6 warning to work from. (api-docs.deepseek.com)