Chinese AI startup DeepSeek has introduced a new model with improved programming abilities to compete with leading US models like OpenAI.
The update, which was released on the open-source AI community platform Hugging Face, the company claims, solves real-world problems and sets new standards for accuracy and efficiency.
DeepSeek’s R1 model, which matched OpenAI’s top offerings, helped it become the most popular free software on Apple’s software Store in January, causing industry astonishment and a selloff in US markets.
DeepSeek is a significant, reasonably priced AI rival because it only invested $6 million to train its V3 model, significantly less than OpenAI’s $100 million for GPT-4, and utilized 10% of the processing power Meta used for Llama 3.1.
Unlike fully open-source AI, DeepSeek’s models include “open weights,” which permit use with certain limitations on modifications.
The company employs top AI scientists from prominent Chinese universities and experts from several industries to boost the intelligence and diversity of its models.
Chinese President Xi Jinping encouraged business executives to realize their full potential during a meeting with entrepreneurs and IT leaders.
“Now is the ideal moment for entrepreneurs and private businesses to realize their full potential,” he told.
