DeepSeek-R1 AI Matches ChatGPT with 2,000 GPUs at $5.58M, Costing 27x Less Than OpenAI

GigaNectar Team

Updated on:

Deepseek user interface. Photo Source : Deepseek

A new AI chatbot from China does what ChatGPT does, but at a fraction of the cost. DeepSeek’s latest AI model, called DeepSeek-R1, matches the performance of leading AI systems while using approximately 2,000 GPUs compared to ChatGPT’s reported 10,000 GPUs.

The key difference? Money and method. DeepSeek built their system for $5.58 million – far less than what companies typically spend on AI development. The model’s operational costs are 27 times less than OpenAI’s o1 model.

“This younger generation also embodies a sense of patriotism, particularly as they navigate US restrictions and choke points in critical hardware and software technologies,” explains Zhang. “Their determination to overcome these barriers reflects not only personal ambition but also a broader commitment to advancing China’s position as a global innovation leader.”

The breakthrough came from an unusual approach. Instead of hiring experienced engineers, DeepSeek founder Liang Wenfeng recruited recent PhD graduates. “Most people, when they are young, can devote themselves completely to a mission without utilitarian considerations,” Liang explained to tech publication 36Kr.

When U.S. export controls limited their access to advanced computer chips, the team found creative solutions. “They optimized their model architecture using a battery of engineering tricks,” says Wendy Chang, an analyst at the Mercator Institute for China Studies. “Custom communication schemes between chips, reducing memory usage, and innovative use of mix-of-models approach – combining these successfully is remarkable.”


Similar Posts


The results speak through numbers. In independent tests, DeepSeek-R1 matched or exceeded the performance of established AI systems from OpenAI, Anthropic, Meta, and Alibaba in tasks like coding, mathematics, and problem-solving.

DeepSeek also made their system “open-weight,” meaning other researchers can examine and improve the underlying algorithm, while the training data remains private. This transparency has earned support from AI researchers worldwide.

At the World Economic Forum in Davos, Switzerland, Microsoft CEO Satya Nadella emphasized, “We should take the developments out of China very, very seriously.”

The impact extends beyond just cheaper AI. DeepSeek’s efficient approach challenges current U.S. export control strategies. As Chang notes, “Existing estimates of how much AI computing power China has, and what they can achieve with it, could be upended.”

Their innovations in efficiency showcase a new direction in AI development. While DeepSeek still needs to prove its real-world applications beyond benchmark tests, their approach suggests advanced AI technology could become increasingly accessible. 

Leave a comment