China’s DeepSeek restructures pricing with a permanent 75% cut on V4-Pro AI model
China’s AI startup DeepSeek has permanently reduced the price of its V4-Pro model by 75%, cutting API costs to between 0.025 and 6 yuan per million tokens, citing structural changes and potential supply advantages from Huawei’s Ascend 950 chips. The move positions DeepSeek competitively amid U.S. trade restrictions limiting Nvidia’s access to China, though Huawei’s production scaling remains constrained by semiconductor export controls.
China’s AI company DeepSeek has introduced a permanent 75% price cut for its flagship V4-Pro model, lowering API costs to between 0.025 and 6 yuan per million tokens depending on usage. The restructuring aligns with broader industry shifts, though DeepSeek did not explicitly confirm whether cost reductions stem from increased supply of Huawei’s Ascend 950 chips, which optimize the V4 model’s performance. The price adjustment positions DeepSeek competitively, as U.S. trade restrictions have blocked Nvidia from selling its most advanced semiconductors in China, benefiting Huawei’s AI chip sales. However, Huawei’s ability to scale Ascend production remains limited due to export controls, which could delay further cost reductions when the Ascend 950 supernodes launch in the second half of 2024. DeepSeek’s V4 model offers two variants: the high-performance Pro version and the more affordable Flash variant. The V4-Pro supports integration with tools like Claude Code, OpenClaw, and OpenCode, enhancing its appeal for developers and enterprises. The pricing shift comes as competing AI models can cost up to 12 times more than weaker alternatives, reinforcing DeepSeek’s focus on affordability. The move may also reflect broader industry trends, where U.S. trade policies indirectly shape China’s AI infrastructure by restricting access to cutting-edge hardware.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.