Chinese AI company DeepSeek plans to unveil its latest language model next week, more than a year after its previous major release, according to information published by the Financial Times and cited by Economedia. The move comes amid intensifying competition between China and US rivals in artificial intelligence.
The Hangzhou-based lab is expected to introduce V4, described as a multimodal model capable of generating images, video and text. According to two sources, DeepSeek has worked with Chinese AI chipmakers Huawei and Cambricon to optimize the new model for their latest domestic processors.
Read also: Anthropic’s Claude AI climbs to No. 2 in the App Store amid dispute with the Pentagon
The strategy reflects China’s broader efforts to reduce reliance on advanced chips produced by Nvidia, which are subject to US export restrictions aimed at curbing Beijing’s technological advancement.
A new milestone for Chinese AI
The launch is scheduled ahead of China’s annual parliamentary meetings, the “Two Sessions,” starting March 4, a high-profile political event that could further elevate DeepSeek’s status as a national AI champion.
This will mark DeepSeek’s first major model release since R1, introduced in January 2025, which the company claimed achieved performance comparable to leading Silicon Valley models while using only a fraction of the computing power.
That announcement sent shockwaves through US tech stocks, with some analysts calling it a “Sputnik moment” for Chinese AI. Since then, DeepSeek has focused on incremental updates, while domestic rivals such as Alibaba and Moonshot gained traction in the open-source AI segment.
Optimizing V4 for China-made chips is expected to boost demand for local semiconductors and accelerate the shift away from Nvidia and AMD processors in inference workloads.
Photo: NBC News
UK
IT
ES
DE
FR
RO
PT
HU
