The Chinese start-up DeepSeek has jolted the tech world with its claim of a powerful AI model that is cheaper to build than better-funded American rivals.

It reportedly matches or surpasses existing AI models in various benchmarks, including programming and reasoning.

It was developed for under US$10 million, significantly less than the billions invested by Silicon Valley companies. And unlike many U.S. tech companies, DeepSeek has made its model open-source, promoting transparency and accessibility. Also, DeepSeek achieved impressive results using fewer and less advanced chips than its competitors, demonstrating innovative approaches to AI development.