OpenAI praises DeepSeek’s affordable AI model

Chinese AI firm DeepSeek has drawn praise from OpenAI for its cost-effective R1 model, raising questions about Big Tech’s billion-dollar AI investments.
OpenAI says powerful AI systems are learning to cheat and hide their intentions.

OpenAI CEO Sam Altman has called Chinese startup DeepSeek’s R1 model “impressive,” highlighting its ability to deliver advanced AI performance at a fraction of the cost. According to DeepSeek, its R1 model is 20 to 50 times cheaper to use than OpenAI’s own models, offering significant affordability without sacrificing quality.

Chinese AI, DeepSeek gained global recognition last month when it revealed that training its DeepSeek-V3 model required less than $6 million in computing resources, leveraging lower-cost Nvidia H800 chips. In contrast, Altman noted that OpenAI remains committed to prioritising increased computing power, suggesting this as an important factor in achieving AI progress.

The emergence of DeepSeek has disrupted the AI industry, leading to a significant sell-off in tech stocks, including Nvidia, which recorded a historic single-day loss of $593 billion in market value. Analysts say DeepSeek’s cost-efficient approach raises doubts about the necessity of the massive financial investments made by US tech firms in AI development.

As DeepSeek continues to attract attention, the startup’s success underscores a shift in the AI market, with low-cost models challenging traditional notions of progress in AI.