China's DeepSeek launches App Store-topping AI model, rivaling OpenAI with fewer resources

DeepSeek's AI model rivals OpenAI with fewer resources, tapping the top of the App Store.

: Chinese startup DeepSeek launched its AI model, R1, which rivals OpenAI's model using fewer resources amidst U.S. chip sanctions. R1, with 671 billion parameters, trained using 2,048 Nvidia H800s and $5.6 million, tops benchmarks such as AIME and MATH-500. Despite geopolitically sensitive challenges, the app quickly became the top free app in the Apple App Store. Founder Liang Wenfeng's unique background in finance and AI helped navigate hardware limitations.

DeepSeek, a Chinese AI startup, has introduced its R1 model, competing against OpenAI's offerings while using significantly fewer resources. This accomplishment is noteworthy due to the restrictions imposed by U.S. sanctions, yet R1 surpasses benchmarks including AIME and MATH-500.

The AI model was developed with 671 billion parameters and was trained on 2,048 Nvidia H800s with a budget of $5.6 million. The release has brought substantial attention both for its performance and geopolitical implications in the field of AI politics.

After its launch, R1 became the top free app in Apple's App Store, outpacing ChatGPT. DeepSeek's founder Liang Wenfeng, known for his work in quantitative finance, successfully navigated the challenges posed by chip restrictions, showcasing adaptability and innovation.