In artificial intelligence, cost-effectiveness is replacing efficiency as the tool to rule the technological landscape. China has laid the foundation of this competition.
DeepSeek is aggressively cutting prices for its recently released flagship V4 model, intensifying competition in China’s tech industry trying to take on Silicon Valley.
In a new value-based change, the AI company is offering a 75 percent discount to developers who are using its newly unveiled DeepSeek-V4-Pro until May 5, 2026. Additionally, the company is reducing the fee for input cash spread across its family of AI platforms to 1/10 of the original price.
The announcement made on X will significantly reduce prices for users who are using these AI models.
Is China waging an AI price war?
DeepSeek’s recent move to cut prices could be seen as a push to revive the price-based war, as last year a Chinese company shocked the global tech industry by releasing the R1.
Other tech giants including OpenAI, Google and Anthropic are also competing with each other to release cutting-edge versions of AI models. Unlike DeepSeek-v4-Pro, access to those models can be expensive for developers and users.
“Pricing, open source availability and a 1 million reference window are all low barriers to entry for developers, startups and small enterprises,” said Akshar Keremane, co-founder of Bengaluru-based AI startup o-Health.
“This allows users to experiment at a modeling capacity and scale that was not available before,” said the entrepreneur, whose Gates Foundation-backed startup deploys AI in large hospitals and rural clinics.
The low-cost model will also encourage users to move toward Chinese AI models, giving Beijing an edge not only in the crowded global AI landscape but also in the competitive US-China race.
The DeepSeek-V4 model was released on Friday. There are two variants of the V4 model: the more powerful Pro with a higher price range and the cheaper Flash variant.
What makes the DeepSeek-v4-Pro model different from other open-source models is its context window. The model can significantly process complex code bases, making it highly suitable for agentic workflows. Additionally, V4-Pro allows easy integration of Cloud Code, OpenClave and OpenCode.
