DeepSeek V4 is set to be released in the next few weeks and may run entirely on Huawei chips

BlockBeats News, April 4th. According to the latest report from the U.S. tech media The Information, Chinese AI startup DeepSeek is set to release its new flagship model V4, which will run entirely on Huawei's self-developed chips, seen as a significant milestone in China's drive for semiconductor self-sufficiency.

V4 is expected to be released in the coming weeks and will run entirely on Huawei chips. DeepSeek has been collaborating with Huawei and chip designer Cambricon for months, rewriting the underlying code of the model to achieve compatibility with domestic chips.

Notably, DeepSeek did not grant early testing access to NVIDIA this time for V4, only inviting domestic chip companies to participate in early optimization, breaking industry norms.

Chinese tech giants such as Alibaba, ByteDance, and Tencent have already pre-purchased Huawei's latest Ascend 950PR chip in large quantities, with total orders reaching hundreds of thousands of units, planning to deploy V4 through cloud services and integrate it into their AI applications. The surging demand has driven up the price of this chip by around 20%.

V4 adopts a Mixed Expertise Architecture (MoE), with a total parameter volume of about 10 trillion, activating approximately 37 billion parameters per inference, supporting multi-modal input of text, images, and code while maintaining low latency.

Last year, DeepSeek released the low-cost models V3 and R1, which caused a global tech stock market plunge, leading the market to question whether U.S. AI companies need to spend billions to purchase computing power. As a result, V4 has attracted international attention.