DeepSeek unveils AI model for domestic chips in symbolic break from Nvidia reliance
DeepSeek released its DeepSeek-V4 AI model series, designed to work with domestic Chinese chips, marking a shift away from reliance on Nvidia's technology. The move is seen as a significant step in China's efforts to build a self-sufficient AI stack amid tightening US export controls on advanced semiconductors.
Chinese AI startup DeepSeek released a new generation of large language models, the DeepSeek-V4 series, explicitly tuned for domestic chips. The models include a high-performance Pro version and a lightweight Flash variant. DeepSeek's technical report validated its models on both Nvidia GPUs and Huawei Ascend NPUs, demonstrating compatibility with domestic hardware. The V4 model supports ultra-long context windows and shows improvements in key AI benchmarks. DeepSeek has completed inference adaptation on Huawei's Ascend platform, moving towards practical deployment on domestic chips. The Beijing Academy of Artificial Intelligence has adapted DeepSeek-V4-Flash for full inference deployment across multiple AI chip architectures.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.