Spending data from Ramp shows Anthropic's cheaper Opus 5 passed the more advanced Fable 5 in corporate spending within a month of its 24 July launch. Anthropic prices Opus 5 at half Fable's rates and says it matches or beats the flagship on bounded coding and office tasks. The readings are contested, as Vercel's separate gateway sample put Fable second in spending, and neither sample covers the whole market.
What changed
Buyers routinely defaulted to the most advanced flagship model for corporate AI work.
What it unlocks
Routing routine tasks to a cheaper model and reserving the flagship for long, self-directed projects.
- $5/M input, $25/M output for Opus 5
- half Fable's rates
- Fable 13.2% of Vercel gateway spending
- avg price per token fell 13.6%
What you need to act on it
- API or gateway access to switch models
Sources