DeepSeek previews new AI model adapted to run on Huawei chips
DeepSeek, a Chinese AI startup, launched a preview of its new AI model, V4, adapted for Huawei chip technology, outperforming other open-source models. The collaboration with Huawei underscores China's growing self-sufficiency in AI, amid U.S. export controls on advanced AI chips.
DeepSeek, a Chinese AI startup, has launched a preview of its new AI model, V4, adapted for Huawei chip technology. The Pro version of V4 outperforms other open-source models in world-knowledge benchmarks, trailing only Google's Gemini-Pro-3.1. DeepSeek collaborated closely with Huawei to ensure V4 can run across its full line of high-performance Ascend systems. Huawei's chips were used in some of V4's training process. The development underscores China's growing self-sufficiency in AI, amid U.S. export controls on advanced AI chips made by Nvidia. Nvidia CEO Jensen Huang expressed concerns that DeepSeek's success on Huawei chips is a 'horrible outcome' for the U.S. Chinese chipmakers rallied on expectations for wider use of homegrown chips. DeepSeek's V4 is suited for AI agent work, which requires more computing power than chatbots.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.