Chinese artificial intelligence (AI) startup DeepSeek announced on Thursday that it plans to comprehensively raise the prices of its API (Application Programming Interface) services soon, with a significant increase expected, according to Yahoo and company sources.
Developers and enterprise customers are advised to manage their usage accordingly. Specific pricing plans and effective dates will be subject to official notification. DeepSeek's announcement of the price increase comes at a time of rapid growth in V4 Flash usage. Statistics from OpenRouter, a global AI model aggregation platform, show that from July 27 to August 2, V4 Flash processed 7.22 trillion tokens in a single week, ranking first in global model call volume on the platform and accounting for approximately 12.7% of OpenRouter's total 56.8 trillion tokens that week, reflecting a significant increase in developer adoption.
V4 Flash currently uses its low price as its main competitive advantage, charging $0.14 and $0.28 per million input and output tokens respectively, far lower than OpenAI, Anthropic, and most international competitors. Research firm Artificial Analysis estimates that the average cost of completing a benchmark test with V4 Flash is only about $0.03, making it one of the lowest-cost models among globally renowned ones.
Top-ranking call volume and increased user stickiness provide DeepSeek with the confidence to shift its competitive strategy from low-price market dominance to revenue monetization. Price increases can improve the gross margin of inference services and help cope with the server and computing power pressure brought about by surging traffic; however, if the price increase is too large, developers may switch to other open-source or lower-priced models. Therefore, the difference between the official price and that of competitors will determine whether DeepSeek can maintain its leading position in global call volume while increasing revenue.
Source:
