On Friday, Chinese startup DeepSeek released a preview of its highly anticipated V4 model, designed to run on chips from Huawei Technologies—marking a shift away from its earlier reliance on Nvidia Corp (NASDAQ:NVDA) hardware.
Performance Claims Put V4 Among Top AI Models
DeepSeek said its V4 Pro model outperforms most open-source competitors on world knowledge benchmarks, trailing only closed-source systems like Alphabet Inc.'s (NASDAQ:GOOG) (NASDAQ:GOOGL) Google's Gemini-Pro-3.1, Reuters reported.
The model is optimized for handling long and complex text tasks, processing more than one million tokens while using significantly less compute than rivals.
However, it lacks multimodal capabilities such as image and video processing.
Huawei Partnership Raises Stakes For Nvidia
The collaboration with Huawei, whose Ascend chips are seen as China's top domestic alternative, signals growing confidence in local hardware.
Previously, Nvidia CEO Jensen Huang warned that such developments could weaken the company's developer ecosystem in China.
U.S. export controls and geopolitical tensions have accelerated China's efforts to build a self-reliant AI stack spanning chips and software.
Risks, Scrutiny And What Comes Next
Despite its rapid rise, DeepSeek faces scrutiny from Western governments over data security concerns, with some restricting its use.
Analysts caution that V4's performance assertions require independent validation. Still, the model's cost efficiency and capabilities suggest open-source AI is narrowing the gap with proprietary systems.
DeepSeek, backed by High-Flyer Capital Management, is reportedly seeking funding at a valuation exceeding $20 billion.
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
Image via Shutterstock
Login to comment