Value AI Models Reshape Spending, Frontier Sales Wobble

Value-for-money AI models are now impacting sales of high-priced, cutting-edge models from companies like Anthropic and OpenAI. A new analysis of spending data from 70,000 companies shows that Anthropic's top-tier Fable 5 model accounted for only about 11% of spending on their tools, more than two months after its launch. What's interesting is this differs from past trends where companies typically chose the best-performing models by default. Analysts suggest the high price of Fable 5 and the belief that existing models meet most business needs are driving this shift. This situation could change the business models for frontier AI developers who have invested heavily in sophisticated models. A venture capitalist, Miles Clements, noted that most users don't need "frontier" capabilities, and the era of customers only choosing these models was unsustainable. Breakthrough technological advances are still vital for ambitious AI promises like treating disease, but these achievements may increasingly serve as a "showcase." This trend is also linked to the rapid evolution of open-weights models, which can be downloaded and modified freely. Even in the US, moves towards open-weights are accelerating. Nvidia, for example, is reportedly developing Nemotron 4, an open-source-based large language model with at least one trillion parameters. This could rank among the top-tier open-source models globally. The bottom line is that the AI market is shifting, with value and open-source options gaining ground against the most expensive, cutting-edge models.