Anthropic is making its most powerful AI coding model cheaper to run

Chinese AI models are gaining traction by being up to 90% cheaper than their US counterparts, potentially disrupting the open market dominated by Google, OpenAI, and Anthropic, according to a Juniper Research report. The study highlights how these Chinese models are not only more cost-effective but also better suited for local deployment, impacting where developers choose to run their AI workloads.

In the past year, around 70% of work conducted on OpenRouter, an API platform and marketplace, utilized US AI models from the aforementioned companies, but this share has now dropped to approximately 30%. The research identifies two distinct AI spending markets: a price-driven one favoring cheaper inference and a quality-driven one where customers are willing to pay more for higher accuracy and reasoning.

Despite US companies still leading the quality-driven market, Chinese models are excelling in the price-driven sector. Moreover, open-weight models are challenging the prevailing notion that inference must be processed in the cloud, potentially diminishing the revenue currently earned by cloud providers. While the future remains uncertain, it is evident that US dominance in the AI industry is being challenged by the rise of affordable Chinese models.