Chinese AI may not be cheaper: Western models cost less for some tasks, study finds

The prevailing notion that artificial intelligence models from China are significantly cheaper may be inaccurate, according to new research. A study indicates that leading Western AI developers, including OpenAI and Anthropic, offer models that are actually less expensive for certain tasks compared to some Chinese open-weight AI models.

The research, conducted by AlphaSense, an AI-driven market intelligence company, found that OpenAI and Anthropic's closed-model AIs provided better responses at a lower cost in some instances than their Chinese open-weight counterparts. This challenges the widespread assumption that Chinese AI development inherently leads to lower operational expenses.

This development could have significant implications for the global AI landscape, potentially shifting perceptions of cost-effectiveness and competitiveness between Western and Chinese AI providers. The study focused on specific tasks to compare the real-world operational costs and performance of these different AI models.