synopsis

The upgraded reasoning model’s capacity was just behind Sam Altman-led OpenAI’s o4-mini and o3 reasoning models on LiveCodeBench.

Chinese artificial intelligence startup DeepSeek has quietly upgraded its RI reasoning model as competition among large-language models heats up.

DeepSeek did not announce the upgrade, but CNBC reported that it was released on the artificial intelligence (AI) model repository Hugging Face.

The report added that the upgraded reasoning model’s capacity was just behind Sam Altman-led OpenAI’s (OPENAI) o4-mini and o3 reasoning models on LiveCodeBench, a site that benchmarks models against different metrics. 

Like the latest upgrade, DeepSeek launched its first large-language model (LLM) iteration, the R1 model, without much fanfare in late January. 

However, the launch sent U.S. AI stocks tumbling, triggering a market crash. Nvidia (NVDA), the biggest AI stock, lost a whopping $600 billion in value, while peers such as ASML (ASML) and AMD (AMD) also came under significant selling pressure.

The DeepSeek model's unique selling proposition is its low cost, even while matching the performance of bigger rivals.

DeepSeek’s rise to prominence comes as the U.S. has imposed more stringent restrictions on AI chip exports to China, and geopolitical tensions between the countries remain on slippery ground amid President Donald Trump’s tariff war.

On Wednesday, Nvidia CEO Jensen Huang criticized the U.S. for restricting AI chip exports to China. On the company’s earnings call, the CEO said, “On export control, China is one of the world's largest AI markets and a springboard to global success.”

“With half of the world's AI researchers based there, the platform that wins China is positioned to lead globally.”

He lamented that the $50 billion Chinese market is effectively closed to U.S. industry. 

Nvidia also flagged more hit from the China chip curb, quantifying its impact on July quarter results at $8 billion.

The Global X Artificial Intelligence & Technology ETF (AIQ), an exchange-traded fund tracking AI and machine learning stocks, has gained about 5% this year.

For updates and corrections, email newsroom[at]stocktwits[dot]com.<