Technical Details
Chinese AI startup DeepSeek permanently reduced prices for its flagship V4-Pro model by 75%, maintaining it at 25% of the original cost. The company did not confirm if this relates to increased Huawei Ascend 950 chip supplies but stated the API cost now ranges from 0.025–6 RMB per million tokens (~$0.0035–$0.83), down from 0.1–24 RMB. A token is a unit of text processed by AI models.
Context and Background
The price cut became possible due to Huawei's chip exports, which benefited from US sanctions against Nvidia. While production equipment export restrictions limited Huawei's Ascend 950 manufacturing capabilities, the company increased chip deliveries. Last month, DeepSeek claimed the Pro version would cost 12 times more than the less powerful Flash due to computational limitations.
Industry Impact
Prices for the Pro version are expected to drop sharply after Huawei launches Ascend 950 super-nodes in the second half of the year. This will strengthen Chinese companies amid US sanctions and chip supply restrictions. Lower AI model prices could stimulate market growth and increase technology accessibility for businesses.