DeepSeek has released a powerful new open-source AI model called V3–0324, which rivals top Western models while running efficiently on local hardware like the Mac Studio.
The model uses a mixture-of-experts architecture with 671 billion parameters, supports 128K context length, and is now fully available under the permissive MIT license.
This major breakthrough in Chinese AI development is drawing global attention, fueling tech rivalry, and reshaping the landscape for startups, governments, and even military applications.