On August 6th, DeepSeek issued a notice stating it plans to significantly raise the overall pricing for its API services in the near future, advising users to "plan your usage accordingly." The specific details will be provided in a formal notice.

This announcement comes just about three weeks after DeepSeek first introduced a peak-valley differential pricing mechanism in mid-July. The shift from peak-load diversion to an overall price increase suggests the company's pricing logic has moved from "managing computational load" towards a comprehensive re-anchoring of its value proposition. The notice did not specify the exact magnitude or timeline of the price adjustment, only urging users to "plan your usage accordingly." For downstream developers who rely on DeepSeek's affordable API to build applications, this means their cost models will need to be re-evaluated.

In mid-July, DeepSeek first implemented a peak-valley pricing mechanism for its V4 series: API prices during peak hours (9:00-12:00 and 14:00-18:00 daily) were set at double the off-peak rates. The V4 Pro model was priced at $0.87 per million output tokens during regular hours, while V4 Flash was listed at $0.28. Goldman Sachs estimated at the time that the blended average price after introducing the peak-valley mechanism was approximately $0.35 per million tokens for V4 Pro and $0.12 for V4 Flash. The market largely interpreted the peak-valley pricing as a signal for refined operational management rather than a fundamental shift in pricing direction.

The rapid transition from peak-valley pricing to a full price increase is driven by clear demand-side factors. The open-source AI agent tool OpenCode disclosed that the official version of DeepSeek V4 Flash on its platform now consumes 8 trillion tokens daily—5 trillion from free credits and 3 trillion from paid plans. In comparison, the large model routing platform OpenRouter, which hosts over 400 models, processes approximately 6.6 trillion tokens daily across its entire platform. A single DeepSeek model at one entry point consumes more than the total daily volume of OpenRouter, which accommodates over 400 models.

Data from the Vercel platform further confirms this trend: DeepSeek's token processing volume has risen to the top spot on the platform, with V4 Flash processing approximately 5.3 trillion tokens per week. As agent tasks evolve from "single-query responses" to "continuous calls lasting several hours with dozens of tool interactions," the demand surge driven by extremely low prices is forcing a reset of pricing strategies.

Liang Wenfeng's Pricing Logic Faces Realization

DeepSeek founder Liang Wenfeng elaborated on the company's pricing philosophy during an investor meeting in July. He stated that API pricing is based on a "reasonable profit model" aimed at recouping investment in equipment purchases within ten months. He also noted that user demand is "almost inelastic" at the current price range—"If the price were halved, the difference in token consumption would be negligible."

This assessment provides a direct basis for the price increase: if demand is insensitive to price, raising prices can significantly boost revenue without substantially sacrificing usage volume. Liang Wenfeng also emphasized that the company does not seek to "maximize profits" but rather "earn only a reasonable return."

Upcoming Release of DeepSeek-V4-Pro Official Version

On July 31st, DeepSeek announced the public beta launch of the official version of DeepSeek-V4-Flash API. This marks the long-awaited official release of the DeepSeek V4 series. According to the company, the model structure and size of DeepSeek-V4-Flash-0731 remain consistent with the DeepSeek-V4-Flash-preview, with only the post-training phase redone. However, DeepSeek noted that the public beta is currently limited to the API interface, and the widely used App and web versions will not have access to the latest capabilities for now. The official version of DeepSeek-V4-Pro is expected to be released soon.

Accelerated Commercialization Push

The announcement of the price increase comes as DeepSeek simultaneously advances multiple commercialization milestones. The company's ARR has reached between $400 million and $500 million, with the flagship V4 model achieving a gross margin exceeding 50%. According to reports citing multiple deal sources, DeepSeek has restarted its second financing round, aiming to raise 50 billion yuan at a pre-money valuation of approximately 500 billion yuan, with a target to complete the deal by late August. This financing round was previously paused at the end of July. Earlier, the company completed its first external financing round of approximately 51 billion yuan.

Regarding the competitive landscape, DeepSeek's comprehensive price increase signals an overall upward shift in the industry's pricing benchmark. The specific pricing plan is still pending formal notice, and key variables to watch include whether the increase exceeds market expectations and whether free credits will be adjusted accordingly.