Anthropic’s best AI model struggles to attract users as cheaper tools thrive

Simon Willison read Ramp's July breakdown of model spending as support for the idea that price held back Anthropic's Fable 5 (link post). Ramp's August 12 AI Index article draws a similar conclusion: Fable 5 was a small part of Anthropic tokens bought, despite costing about twice as much as OpenAI's GPT-5.6 Sol.

The Fable figures come from businesses using Ramp's token spend management product, a narrower sample than the headline adoption numbers. Ramp says it skews more technical than its usual sample and that real adoption is probably lower. The price explanation is the authors' interpretation of observed spending, not a controlled test of why buyers chose other models.