Find XO: Chinese Open Models Cut Inference Costs by up to 90%, But Entrepreneurs Should Focus on Where Their Data Goes (ft.com)

- As OpenAI and Anthropic API prices rise, U.S. and European companies are switching to Chinese open-weight models like DeepSeek and Q1. - Because they handle the same tasks 60% to 90% cheaper. - However, lower inference costs do not necessarily mean lower total costs. - Korean entrepreneurs should calculate not only inference costs but also where prompts and logs are sent legally, including personal data overseas transfer and regulatory compliance, to realize actual savings.