Anthropic's Fable 5, widely considered the most capable AI model on the market, is struggling to win corporate buyers, according to new spending data from financial services provider Ramp. In its first month after launch, Fable 5 accounted for only about 6% of tokens purchased from Anthropic and 11.4% of total spending on the company's models. The figures suggest a ceiling on what businesses will pay for frontier AI, even when the model leads the pack.
A Slow Start for the Flagship
Fable 5's debut in July 2026 was supposed to cement Anthropic's lead. Instead, Ramp's AI Index shows companies are barely buying the model through its API. By comparison, OpenAI's GPT-5.6 Sol captures 25% of tokens and 23% of spending at OpenAI. Fable 5 brought in only about 75% of the model-related revenue that GPT-5.6 Sol generated.
The price gap is stark. Fable 5 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. Ramp economist Ara Kharazian attributes Fable 5's slow uptake to its price. "The extra performance of Fable 5 simply isn't worth the cost," he said.
Kharazian sees this as a new ceiling on what companies are willing to spend on AI. The performance edge Fable 5 offers may not matter for many use cases, or it's barely measurable in daily work. Calculating AI return on investment is complicated and messy, and for most firms, the math doesn't favor the premium.
Adoption Shifts Among U.S. Companies
Ramp's data, which skews slightly toward tech companies, shows Anthropic has passed OpenAI in adoption among U.S. companies and is widening its lead. In July, 43.5% of U.S. companies paid for Anthropic subscriptions or tokens, up 1.1 percentage points from the previous month. OpenAI reached 39.7% adoption but grew by only 0.23 percentage points.
xAI posted its fastest growth since July 2025, rising 0.94 percentage points to 4%. That pace stands out in a market where growth at OpenAI and Anthropic is decelerating. New customers keep signing up with American model providers, but the expansion is slowing.
Actual Fable 5 adoption is likely even lower than estimates, assuming the model is used mainly for coding. Ramp's sample skews slightly toward tech companies, which are the most likely to test new coding models. If those firms aren't buying, the broader corporate market is probably even more cautious.
The Open-Source Challenge
Advanced users are shifting toward open-source models, which now trail frontier models by only a few months. That narrow gap erodes the case for paying a premium. Companies will buy what pays off, but as long as value stays abstract, willingness to pay appears limited.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
The data doesn't say that a Fable 5 class model represents the upper limit of what companies would pay for AI per se. Models that are dramatically more capable could deliver dramatically higher and more tangible value. But for now, the willingness to pay sharply higher prices for performance gains that are difficult to measure appears to have plateaued.
Ramp says these are worrying signs for the AI industry. The industry's investment thesis depends on fast-growing revenue from increasingly powerful models. If corporate buyers balk at the price of the best model, that thesis weakens.
Spending Rises, But Not Without Limits
Total AI spending keeps rising. In July, the top 1% of U.S. companies spent a median of $7,400 per employee on AI. For the top 10%, that figure was $650. The median company spent $11.95 per employee.
Those numbers show a market that is still growing, but unevenly. The heaviest spenders are investing heavily, while the typical firm commits only a modest amount. Ramp's data suggests companies are spending more on AI, but not without limits.
The gap between the top 1% and the median company is enormous, and it reflects a divide in how AI is being used. Large tech-heavy firms can absorb the cost of frontier models like Fable 5. Smaller companies, facing the same $10 and $50 per million token prices, are far more hesitant.
What This Means for the AI Industry
Kharazian's analysis points to a simple conclusion: price matters, and performance alone doesn't justify it. Fable 5's slow adoption is a signal that even the most capable model can't command unlimited corporate budgets.
The article, published Aug 13, 2026, draws on Ramp's proprietary token spend management product. It is part of The Decoder's "AI in practice" section, and it was prompted by Nano Banana Pro, an AI tool.
The broader picture is mixed. Adoption is still rising, and total spending is climbing. But the pace is slowing, and open-source alternatives are closing the gap. For Anthropic, the challenge is clear: Fable 5 may be the best model on paper, but in the market, it's facing a hard ceiling.