According to a recent report from the investment bank Jefferies, the financial barrier for enterprises to utilize artificial intelligence has significantly decreased. Between August 6 and August 8, the average cost for AI inference—the process of a model handling data—fell to a range of US$1.16 to US$1.18 per million tokens. This figure represents the lowest recorded cost for the year to date.

Market analysts point to two primary drivers behind this downward trend. First, a global price war among major AI developers has forced companies to lower their service fees to remain competitive. Second, there has been a notable surge in the adoption of low-cost, open-source AI tools. Specifically, the report highlights the influence of Chinese-developed models, such as those produced by DeepSeek, which are providing businesses with cheaper alternatives to established proprietary systems. By integrating these open-source options, companies are effectively reducing their operational overhead while maintaining access to advanced machine learning capabilities. This shift suggests that the market is moving toward a more commoditized landscape where accessibility and cost-efficiency are becoming as critical as the performance of the models themselves.