Two months after release the flagship sits third in its own lineup on 8.0%, in card-billing data from the 70,000 companies that use Ramp's corporate cards.
Fable 5 has been available too long for a slow start to explain the gap: Opus 4.8, a generation older, still draws three and a half times as much of what businesses spend on Anthropic's models, on Ramp's figures.
Ramp's July breakdown gives Opus 4.8 28.0% of the spending, Sonnet 4.6 8.3% and Fable 5 8.0%, with Opus 4.6 on 6.9% and the newer Opus 5, released on 24 July, on 3.5%. Cost, not capability, is the explanation attached to that ordering: Fable 5 is priced steeply enough, the argument runs, to have dented its own uptake.
Developer Drew Breunig has described what a price like that does inside an engineering team. Before Fable arrived, he wrote, it felt pointless to spend much time optimizing a coding harness or a context strategy, because a new model would typically turn up soon at the same price or a lower one and paper over most of those problems.
Fable was, and still is, an excellent model, he wrote. But its price was steep enough that other models — among them Opus, K3 and even GLM — did the job well enough for most of the coding he needed, and that pushed him into weighing which model to hand which task.
A top-priced tier is not automatically a hard sell. Sol costs more than anything else OpenAI sells and is reported to be far and away the most used model in its lineup, which suggests the problem is price set against how much of the work a cheaper model already does, rather than price on its own.