DeepSeek Releases V4 AI Model, Huawei Pledges $11.7 Billion for Autonomous Driving

Chinese AI startup DeepSeek released preview versions of its V4-Pro and V4-Flash foundational models on Friday, with the Hangzhou-based company claiming V4-Pro beats all rival open models for maths and coding while trailing only Google's Gemini 3.1-Pro for world knowledge. The release follows DeepSeek-R1's January launch, which the company claimed cost less than $6 million to develop compared to Silicon Valley's multibillion-dollar budgets, though some tech analysts dispute this account and argue DeepSeek had access to greater resources than acknowledged. The development prompted Marc Andreessen to call it AI's Sputnik moment, while multiple governments including US states, Australia, Taiwan, South Korea, Denmark, and Italy introduced bans or restrictions citing privacy and national security concerns. Separately, Huawei announced plans to invest 80 billion yuan ($11.7 billion) over five years in autopilot training systems, underscoring China's major push into AI and autonomous driving technologies.

DeepSeek-V4 model release and performance capabilities

• ▪DeepSeek-V4-Pro's performance trails state-of-the-art frontier models by approximately 3 to 6 months according to the Hangzhou-based startup
• ▪DeepSeek-V4-Pro and DeepSeek-V4-Flash follow an open-source model, allowing developers to use and modify the source code freely
• ▪DeepSeek-V4-Pro beats all rival open models for maths and coding according to DeepSeek
• ▪DeepSeek launched preview versions of DeepSeek-V4-Pro and DeepSeek-V4-Flash on Friday
• ▪DeepSeek-R1 launched in January of the previous year with capabilities broadly comparable to ChatGPT and Gemini
• ▪DeepSeek-V4-Pro trails only Google's Gemini 3.1-Pro for world knowledge among all models according to DeepSeek
• ▪DeepSeek-V4-Flash has similar reasoning abilities to DeepSeek-V4-Pro while offering faster response times

DeepSeek's cost-efficiency claims and industry skepticism

• ▪DeepSeek-R1's claimed computing costs of less than $6 million were a fraction of the multibillion-dollar budgets usual in Silicon Valley
• ▪Marc Andreessen hailed DeepSeek-R1's release as AI's Sputnik moment
• ▪Marc Andreessen is a prominent Silicon Valley venture capitalist with close ties to United States President Donald Trump
• ▪DeepSeek-R1 developers claimed to have spent less than $6 million on computing costs

Global regulatory response and security concerns

• ▪South Korea introduced bans or restrictions on DeepSeek-R1 shortly after its release citing privacy and national security concerns

Perspective of DeepSeek

• ▪DeepSeek-R1 developers claimed their model achieved capabilities broadly comparable to ChatGPT and Gemini while spending less than $6 million on computing costs

Perspective of Governments concerned about DeepSeek

• ▪Multiple US states cited privacy concerns as justification for introducing bans or restrictions on DeepSeek-R1

Perspective of Marc Andreessen and Silicon Valley supporters

• ▪Marc Andreessen characterized DeepSeek-R1's release as AI's Sputnik moment, suggesting it represents a major competitive challenge