Chinese AI startup DeepSeek announced a permanent 75% reduction in pricing for its flagship V4-Pro model, significantly lowering the cost of accessing advanced AI capabilities. This move follows the initial launch last month when higher compute costs had kept prices much steeper.
- V4-Pro AI model price cut to 25% of original cost
- Pricing now ranges from 0.025 to 6 yuan per million tokens
- Huawei Ascend 950 chip availability likely supports cost reduction
What happened
Chinese AI startup DeepSeek has announced a permanent price cut of 75% on its flagship V4-Pro artificial intelligence model. The company reduced API prices to between 0.025 and 6 yuan per million tokens, a substantial decrease from the previous range of 0.1 to 24 yuan per million tokens. This reduction means users can now access advanced AI capabilities at a quarter of the previous cost, improving affordability significantly.
DeepSeek initially launched the V4-Pro model last month with pricing approximately 12 times higher than its less powerful Flash version, attributing the difference to limited access to high-end compute resources. The new reduced pricing is said to be permanent and reflects improvements in hardware supply chains and production capacity.
Why it matters
The dramatic price cut comes as Huawei’s Ascend 950 AI chips, which DeepSeek uses to maximize the V4-Pro model’s performance, become more available. While U.S. export controls have restricted Nvidia’s ability to supply advanced semiconductors in China, they have simultaneously benefitted Huawei’s chip sales. However, limitations on chipmaking equipment exports have constrained Huawei’s capacity to scale production quickly.
Lowering the cost of AI model usage could accelerate adoption and innovation in AI-driven industries, especially in major markets like China and India. With a more accessible pricing structure, startups, developers, and enterprises may increasingly integrate DeepSeek’s powerful AI tools into their applications, potentially driving competitive dynamics in the regional AI landscape.
What to watch next
Observers will be closely monitoring how Huawei scales the production of its Ascend 950 supernodes in the second half of 2026, as the availability of these chips is expected to be a critical factor in sustaining DeepSeek’s pricing strategy. The supply chain and production capabilities will likely influence whether further price reductions or performance enhancements are forthcoming.
Additionally, attention will focus on how DeepSeek’s pricing adjustments impact market competition in AI services across China and India. The permanent cost reduction may pressure competitors to also lower prices or improve offerings, while enabling a broader segment of users to access cutting-edge AI technologies, potentially reshaping the AI software market landscape in the region.