Fable 5 grabs just 6% of Anthropic token spend — corporate AI budgets have limits
Curated by the Inblix editorial team
Anthropic’s Fable 5 may top the charts for raw capability, but corporate America isn’t reaching for its wallet. Ramp’s spending data shows the model accounted for just six percent of tokens purchased from Anthropic in its first month, and 11.4 percent of total spend on the company’s models. Compare that to GPT-5.6 Sol, which pulls 25 percent of OpenAI tokens and 23 percent of spending. Fable 5 generated roughly 75 percent of the revenue Sol did despite costing about twice as much per token — $10 per million input and $50 per million output.
Ramp economist Ara Kharazian pins the sluggish adoption on price, calling it evidence of a new ceiling on what companies will pay for AI. But the real story is messier. Fable 5’s performance edge over cheaper alternatives is often imperceptible in day-to-day work, and that creates a brutal ROI equation. How do you quantify the value of a model that’s marginally better at tasks where “good enough” already gets the job done? The gap between model generations is shrinking in practical terms, even as sticker prices climb.
There’s a broader signal here about where the market is heading. Ramp’s data shows advanced users — the heavy spenders OpenAI and Anthropic increasingly depend on — are drifting toward open-source models that now trail frontier systems by only a few months. Growth is decelerating at both labs: Anthropic added 1.1 percentage points in July to reach 43.5 percent company penetration, while OpenAI grew just 0.23 points to 39.7 percent. Meanwhile xAI posted its fastest growth since July 2025.
Total AI spending still climbed in July — the top one percent of companies spent a median of $7,400 per employee — but the premium tier is stalling. Companies will pay for models that demonstrably pay off. What they won’t do, at least not indefinitely, is subsidize abstract performance gains. For an industry whose investment thesis depends on fast-growing revenue from ever-more-powerful models, that’s the warning buried in Ramp’s numbers.
💡 Key Takeaways
- Fable 5 captured only 6% of Anthropic token purchases in its first month despite being widely considered the most capable model available
- Ramp economist Ara Kharazian attributes the slow uptake to pricing that runs roughly double GPT-5.6 Sol's cost per token
- Advanced users are shifting toward open-source models that trail frontier systems by only a few months, slowing growth at both OpenAI and Anthropic
- Total AI spending still rose in July, but companies are refusing to pay premium prices for performance gains that are hard to measure in daily work
Keep reading: See related articles below for more coverage on this topic.
Get smarter about AI
The sharpest AI news, curated daily. Delivered free to your inbox.