DeepSeek launches 1.6 trillion parameter V4 on Huawei chips as U.S. escalates AI theft accusations — U.S. gov't alleges IP theft by DeepSeek and other Chinese AI firms

DeepSeek has unveiled its V4 large language model with 1.6 trillion parameters, optimized for Huawei's Ascend AI processors. The release coincides with U.S. accusations of IP theft against DeepSeek and other Chinese AI firms. The V4 model comes in two versions, V4-Pro and V4-Flash, priced at $3.48 and $0.28 per million output tokens respectively. DeepSeek's V4 is said to outperform open-source competitors in coding and reasoning benchmarks, training on Huawei chips to avoid Nvidia hardware restrictions. The U.S. government's allegations of IP theft by Chinese AI firms add to existing claims made by companies like Anthropic and OpenAI.