Fable 5's slow adoption suggests corporate willingness to pay for frontier AI has hit a ceiling
Anthropic's Fable 5, despite being considered the most capable AI model, captured only ~6% of tokens and 11.4% of spending on Anthropic's API in its first month Fable 5 generated only ~75% of the revenue that OpenAI's GPT-5.6 Sol produced, despite costing roughly twice as much per token ($10/$50 per million input/output tokens) Ramp data suggests a pricing ceiling has emerged: companies are unwilling to pay sharply higher prices for marginal performance gains that are difficult to measure in dai
Analysis
TL;DR
- Anthropic's Fable 5, despite being considered the most capable AI model, captured only ~6% of tokens and 11.4% of spending on Anthropic's API in its first month
- Fable 5 generated only ~75% of the revenue that OpenAI's GPT-5.6 Sol produced, despite costing roughly twice as much per token ($10/$50 per million input/output tokens)
- Ramp data suggests a pricing ceiling has emerged: companies are unwilling to pay sharply higher prices for marginal performance gains that are difficult to measure in daily work
- Advanced users are increasingly shifting toward open-source models, which now trail frontier models by only a few months, contributing to decelerating growth at both OpenAI and Anthropic
- Total AI spending continues to rise, but adoption is highly skewed: top 1% of companies spend a median of $7,400/employee vs. $11.95 for the median company
Why It Matters
This data reveals a critical inflection point for the AI industry: the investment thesis of "more capable models → more revenue" is facing headwinds as corporate buyers hit a willingness-to-pay ceiling. For AI practitioners and investors, it signals that frontier model providers can no longer rely purely on performance improvements to drive revenue growth, and must instead demonstrate tangible, measurable ROI to enterprise customers.
Technical Details
- Fable 5 pricing: ~$10 per million input tokens and ~$50 per million output tokens, approximately double the cost of GPT-5.6 Sol and other Anthropic flagship models
- Ramp AI Index methodology: based on proprietary token spend management data, skews toward tech companies, meaning actual Fable 5 adoption is likely even lower than reported
- OpenAI GPT-5.6 Sol captured 25% of tokens and 23% of spending at OpenAI in the same period, significantly outperforming Fable 5's market share
- Anthropic reached 43.5% adoption among U.S. companies in July (up 1.1pp), while OpenAI reached 39.7% (up only 0.23pp), with xAI at 4% (up 0.94pp)
- Spending distribution in July: top 1% of companies spent a median of $7,400/employee on AI, top 10% spent $650/employee, and the median company spent $11.95/employee
Industry Insight
- Frontier model providers should reconsider premium pricing strategies for incremental capability gains; the market appears to have a hard ceiling on what enterprises will pay for performance improvements that lack clear, measurable business impact
- The shift of advanced users toward open-source models poses a structural threat to the subscription/revenue growth model of closed AI labs, as the capability gap narrows to mere months rather than years
- Companies that can demonstrate tangible, quantifiable ROI from their AI tools — rather than relying on benchmark-leading performance — will be better positioned to capture and retain enterprise spending as the market matures
Disclaimer: The above content is generated by AI and is for reference only.