DeepSeek releases V4 AI model that matches US competitors—and works with Chinese chips, reshaping the global AI supply chain

DeepSeek, a Chinese AI startup, previewed its V4 model on Friday, claiming it performs at the same level as closed-source competitors like OpenAI's ChatGPT, Google Gemini, and Anthropic's Claude—but as open-source software anyone can download and modify for free.

V4 shows major gains in coding tasks (writing and fixing software code), which has become essential for AI agents (AI systems that complete multi-step tasks on their own). This matters because code-writing ability directly determines whether AI can automate knowledge work—from data analysis to software development.

For business professionals and startups: V4 works with Huawei chips made in China, not just expensive NVIDIA hardware. This means teams outside the US can now build competitive AI systems without relying on American companies, opening AI development to countries and companies previously locked out by US export restrictions.

V4 is available as open-source—you can download it immediately. This announcement arrives exactly one year after DeepSeek's R1 model shocked US competitors by matching their performance at a fraction of the cost, signaling China is closing the AI capability gap faster than expected.