Models ·
Anthropic’s Fable 5 Has a Price Ceiling Problem
Models ·
Anthropic’s Fable 5 Has a Price Ceiling Problem
AI summary · Generated from this article
Anthropic’s Fable 5 captured limited enterprise spending despite leading benchmarks, as buyers reserved the premium model for tasks that clearly justified its price. In its first full month, Fable 5 represented 11.4% of Anthropic model spending and 6% of Claude tokens among companies using Ramp’s token management product. It costs $10 per million input tokens and $50 per million output tokens, versus $5 and $25 for Claude Opus 5, $2 and $10 for Sonnet 5, and $3 and $15 for open-weight Kimi K3. Adoption may also have been affected by its June 9, 2026 launch, June 12 suspension, July 1 restoration, stricter cyber safeguards, and 30-day business-traffic retention. The data points to workload fragmentation, not an Anthropic collapse: 43.5% of US businesses in Ramp’s July dataset paid for Anthropic, ahead of OpenAI’s 39.7%. The key risk is margin pressure as customers build multi-model stacks and use Fable mainly as an escalation layer.