Anthropic's Fable 5 is described as the most capable AI model on the market, yet early spending data suggests that corporate buyers are not rushing to use it. The signal matters because frontier AI economics depend on companies paying more for stronger models, not just experimenting with AI in general.
Data from financial services provider Ramp shows a split between interest in AI and willingness to pay the highest prices for it. Companies are still spending on AI, but Fable 5's first-month API uptake suggests that even advanced models have to clear a practical return-on-investment test.
Fable 5 is powerful, but adoption is narrow
In its first month after launch, Fable 5 represented only about six percent of the tokens purchased from Anthropic through its API, according to Ramp's spending data. By spending, it accounted for 11.4 percent of total Anthropic model spend.
That is a modest result for a model positioned at the top of the market. OpenAI's flagship model, GPT-5.6 Sol, captured 25 percent of tokens and 23 percent of spending at OpenAI. Ramp also found that Fable 5 generated only about 75 percent of the model-related revenue produced by GPT-5.6 Sol, even though Fable 5 costs significantly more per token.
Ramp notes that the Fable data comes from its proprietary token spend management product and is slightly tilted toward tech companies. The source also says actual adoption of Fable 5 is likely even lower than the estimates if the model is used mainly for coding.
Price is the clearest pressure point
Ramp economist Ara Kharazian points to price as the likely reason for Fable 5's slow start. The model costs about $10 per million input tokens and $50 per million output tokens. That makes it roughly twice as expensive as GPT-5.6 Sol or other Anthropic flagship models.
The pricing creates a simple business question: how much extra output does a company get for the extra cost? Kharazian argues that Fable 5 may be exposing a new ceiling in corporate AI spending because the added performance is not worth the higher price for many buyers.
The source frames the issue as more complicated than price alone. A model can be better in a general sense while still offering too little practical improvement for common workflows. If the difference is hard to see in day-to-day work, it becomes difficult for a company to justify the premium.
The ROI problem is becoming harder to ignore
Fable 5's adoption pattern highlights a broader challenge in AI return on investment. Companies can count tokens and invoices, but measuring the business value of one model generation over another is harder.
That problem becomes sharper when prices rise faster than the benefits buyers can clearly identify. If a company cannot put a reliable number on the value delivered by a stronger model, the model's technical advantage may remain too abstract for budget decisions.
The data does not prove that companies will never pay more for advanced AI. The source makes a narrower point: buyers may pay for models that deliver dramatically higher and tangible value. But when that value is unclear, willingness to pay appears limited.
OpenAI and Anthropic are still growing, but more slowly
Ramp's broader data shows that AI adoption has not stalled. In July, 43.5 percent of U.S. companies paid for Anthropic subscriptions or tokens, up 1.1 percentage points from the previous month. OpenAI reached 39.7 percent, but its growth was only 0.23 percentage points.
xAI grew faster in the same period. It rose 0.94 percentage points to 4 percent, which Ramp described as its fastest growth since July 2025.
The source also points to a shift among advanced users. New customers continue to sign up with American model providers, but users whose rising spending matters to OpenAI and Anthropic are moving toward open-source models. Ramp's data says those models now trail frontier models by only a few months.
Total AI spending is still rising
The caution around Fable 5 does not mean companies are walking away from AI. Ramp's July data shows a wide spread in how much U.S. companies spend per employee on AI.
- The top 1 percent of U.S. companies spent a median of $7,400 per employee on AI.
- The top 10 percent spent a median of $650 per employee.
- The median company spent $11.95 per employee.
Those figures show that the market is still expanding, but unevenly. A small group of companies is spending heavily, while the typical company remains far more cautious.
For the AI industry, the concern is not simply whether companies use AI. It is whether they will keep paying sharply higher prices for increasingly powerful models. Ramp's data suggests the answer depends on whether the performance gains are visible enough, measurable enough and valuable enough to justify the premium.