DeepSeek Unveils Upgrade Of Flagship AI Model With Support For Chinese Chips As Beijing Races To Cut Reliance On Nvidia, US Tech

On Thursday, Chinese AI startup DeepSeek released an upgraded version of its flagship V3 model, touting support for domestic chips as China accelerates efforts to reduce dependence on U.S. companies like Nvidia Corporation NVDA.

DeepSeek's New V3.1 With Domestic Chip Compatibility

DeepSeek said its V3.1 model upgrade features faster processing and a new UE8M0 FP8 precision format optimized for "soon-to-be-released next-generation domestic chips," reported Reuters.

The company did not disclose which manufacturers or chip models would be supported.

FP8, or 8-bit floating point, enables AI models to run more efficiently by using less memory and delivering quicker performance.

DeepSeek described the V3.1 update as incorporating a hybrid inference structure, allowing it to toggle between reasoning and non-reasoning modes.

A new "deep thinking" button has been added to its app and web platform, now running the upgraded model.

DeepSeek also said it will adjust pricing for developers using its API — which integrates its models into third-party applications — starting Sept. 6.

Also Read: Alibaba Chair Says AI Total Addressable Market Is $10 Trillion, Expects Benefits Via Its Cloud Business

Past Struggles With Huawei Hardware

Earlier this month, it was reported that DeepSeek delayed its new R2 model launch due to persistent technical problems with Huawei Technologies's Ascend processors. The startup was forced to fall back on Nvidia chips for training, while relying on Huawei hardware only for inference tasks.

DeepSeek's Growing Impact On Global AI

DeepSeek first made headlines in January when its R1 model triggered a $600 billion sell-off in Nvidia's market value. Founder Liang Wenfeng has since pushed for aggressive model updates, including the V3 release in March and the R1 update in May.

Nvidia acknowledged DeepSeek's early success, saying the startup's work demonstrated how models can scale using widely available compute while remaining compliant with U.S. export controls.

China's Broader Push For Tech Independence

The latest development comes as Beijing pushes for domestic chip adoption. Earlier this week, it was reported that Nvidia has reportedly instructed suppliers to pause production of its H20 AI chip for China, amid pressure from Beijing regulators.

Read Next:

• AMD CEO Lisa Su Says China Strategy Rebounding As MI308 AI Chips Await US License: ‘Better Position Than We Were Ninety Days Ago'

Photo: mundissima on Shutterstock.com

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.