DeepSeek Launches V4 With Huawei Ascend Chips, Cutting Prices To A Fraction Of OpenAI
Chinese AI lab DeepSeek released preview versions of its V4-Pro and V4-Flash models on Friday, with the flagship Pro variant carrying 1.6 trillion total parameters, 49 billion active, a 1-million-token context window and open-source weights on Hugging Face. The lab partnered with Huawei, whose Ascend supernode clusters built on the new 950 chips will run V4 at scale in the second half of the year, marking a decisive pivot away from the Nvidia hardware that powered V3 and R1. DeepSeek priced V4-Pro at $3.48 per million output tokens against OpenAI's $30 and Anthropic's $25, with V4-Flash at just $0.28 per million.
Why it matters for Asia
The V4 launch is the clearest signal yet that China's AI stack, model plus silicon, is now a coherent alternative rather than a lagging copy. For enterprise buyers across Asia, it reframes the procurement calculus - sovereign deployments on Huawei hardware running best-in-class open weights become viable in Singapore, Jakarta and Riyadh, where export controls or data residency rules had made US cloud AI awkward. Expect a fresh round of price compression from Alibaba, Tencent and Baidu within weeks. Read more: [CNBC](https://www.cnbc.com/2026/04/24/deepseek-v4-llm-preview-open-source-ai-competition-china.html)