BEIJING: Chinese artificial intelligence (AI) startup DeepSeek has introduced a discounted off-peak pricing model for developers using its AI models, a move that industry analysts suggest could pressure rivals to reconsider their pricing strategies.
According to DeepSeek’s website, the company announced that API usage fees would be reduced by up to 75% between 16:30 GMT and 00:30 GMT.
The R1 model API will be available at a 75% discount, while the V3 model API will be offered at 50% off during this period.
DeepSeek refers to this period as “off-peak” based on Beijing time (00:30–08:30), though it coincides with regular daytime hours in Europe and the United States.
The discounted rates are expected to appeal to developers in these regions, potentially pressuring competitors like OpenAI and Google to introduce similar cost-saving measures.
Wei Chen, a technology analyst at Asia Tech Insights, described the strategy as “a classic example of applying traditional utility pricing models to cutting-edge technology.”
He noted that by offering lower prices during off-peak hours, DeepSeek is optimising its infrastructure efficiency while making its AI models more accessible to cost-conscious developers and startups.
Competition in AI industry
DeepSeek’s pricing move comes at a time of intensifying competition in the AI industry. Since launching its AI assistant last month, OpenAI, the maker of ChatGPT, has reduced prices, while Google’s Gemini has introduced discounted access tiers.
Industry observers believe DeepSeek’s aggressive pricing model could further shake up the market.
The Hangzhou-based company has gained significant attention since the release of its R1 model in January, which reportedly triggered a sell-off exceeding US$1 trillion in global equity markets as investors worried about its impact on existing AI leaders.
DeepSeek’s cost-effective AI models have already led to AI price war, particularly following the release of its V2 model last May.
The company is now reportedly accelerating the launch of the R2 model, the successor to R1, according to sources familiar with its plans.
Sarah Jameson, Director of the Institute for AI Economics, highlighted the broader implications of the move. “If successful, this could create a ripple effect across the AI sector. Major players like OpenAI and Anthropic may need to revisit their pricing models, particularly if DeepSeek attracts a significant number of developers.”
AI industry’s economic model
DeepSeek’s pricing structure resembles traditional off-peak utility pricing, offering lower costs during periods of reduced demand.
This model could be particularly beneficial for startups and independent developers who often face financial constraints when accessing high-end AI models.
ALSO READ: AI Has to Do “100 Times More” Computation After ChatGPT: Nvidia CEO
The company’s unconventional approach is attributed to its founder, Liang Wenfeng, a billionaire who made his fortune through the quantitative hedge fund High-Flyer.
DeepSeek operates more like a research-driven entity rather than a traditional profit-focused enterprise, contrasting with the hierarchical management style typical of China’s tech industry.
Vijayasimha Alilughatta, Chief Operating Officer of Indian tech services provider Zensar, noted that DeepSeek’s cost-effective AI models challenge conventional industry economics.
“Their ability to create competitive AI models at significantly lower costs is breaking the stranglehold of dominant players in the sector,” he said.
