DeepSeek is said to be currently working on its next-gen AI model called V4, and this model is rumoured to be released in the coming weeks. Ahead of its rumoured release, folks at The Information have revealed that the upcoming V4 model will run on the latest chips designed by Huawei Technologies, US.
It is said that the V4 model will only and completely use the latest Huawei AI chips, while in preparation for V4’s launch, Chinese tech giants, including Alibaba Group, ByteDance and Tencent Holdings, are said to have placed bulk orders for Huawei’s upcoming chip totalling hundreds of thousands of units.
Reportedly V4 AI model could use a nearly 1-trillion parameter architecture with probably hundreds of thousands of Huawei’s chips to trigger 1.8x faster inference speeds, a 1-million-token context window, and enhanced efficiency via Engram.
DeepSeek is said to have spent the past few months working directly with Huawei and another Chinese chip designer, Cambricon Technologies, to help rewrite pieces of the model’s underlying code and in testing.
The report also mentions that DeepSeek is working on two additional V4 variants, each optimised for different capabilities and built to run on Chinese chips. As of the writing of this article, neither DeepSeek nor Huawei has made any announcement regarding this latest collaboration. But since in the past DeepSeek has used Huawei’s Ascend chips in its AI models, there are high chances that the above information is true.
