Anthropic's Fable 5 Captures Only 11.4% of Anthropic Spending, Signaling Price Ceiling for Frontier AI

Slow Uptake for Anthropic's Most Capable Model

Anthropic's Fable 5, widely regarded as the most capable model on the market, is seeing weak adoption among U.S. businesses despite its performance reputation. Spending data from financial services provider Ramp shows Fable 5 accounted for only about 6% of tokens purchased from Anthropic in its first month after launch, and 11.4% of total dollar spending on Anthropic models.

By comparison, OpenAI's flagship model, GPT-5.6 Sol, captures 25% of tokens purchased from OpenAI and 23% of OpenAI's model spending. According to Ramp, Fable 5 generated only about 75% of the model-related revenue that GPT-5.6 Sol brought in, despite costing significantly more per token.

Ramp notes its data comes from its proprietary token spend management product and skews slightly toward tech companies. Actual Fable 5 adoption is likely lower than these figures suggest, since the model is believed to be used primarily for coding tasks.

Pricing May Be Hitting a Wall

Fable 5 costs approximately $10 per million input tokens and $50 per million output tokens — roughly double the price of GPT-5.6 Sol or other Anthropic flagship models. Ramp economist Ara Kharazian attributes the slow adoption directly to this pricing gap, arguing companies don't see enough additional value to justify the premium.

The underlying issue may be more complex than price alone. Fable 5's performance advantages may not be significant for many real-world use cases, or the improvement may be difficult to detect in daily work. This points to a broader measurement problem in enterprise AI: companies struggle to quantify the return on investment from incremental model upgrades, particularly when comparing one generation to the next.

The data doesn't necessarily indicate that Fable 5-level pricing represents an absolute ceiling for AI spending. A model offering dramatically higher, more tangible capability gains could command higher prices. But as long as performance improvements remain abstract or hard to measure, corporate willingness to pay appears constrained.

Growth Slows for Both OpenAI and Anthropic

According to Ramp, 43.5% of U.S. companies paid for Anthropic subscriptions or API tokens in July, up 1.1 percentage points month-over-month. OpenAI reached 39.7% adoption but grew only 0.23 percentage points — lagging broader AI adoption trends. xAI posted its fastest growth since July 2025, rising 0.94 percentage points to reach 4% adoption.

While new customers continue signing up with major American providers, Ramp's data suggests high-spending advanced users — the group OpenAI and Anthropic increasingly depend on for revenue growth — are shifting toward open-source models. Open-source models now trail frontier models by only a few months in capability, according to Ramp, contributing to the deceleration at both leading labs.

Overall AI spending continues to climb. In July, the top 1% of U.S. companies by AI spend paid a median of $7,400 per employee. The top 10% spent a median of $650 per employee, while the median company spent just $11.95 per employee — highlighting a steep spending divide.

What This Means

Ramp's data suggests the AI industry's revenue growth model — betting on customers paying steadily more for incrementally more capable models — is running into resistance. Companies aren't rejecting AI spending outright; total spend keeps rising. But they are increasingly skeptical of paying double for performance gains they can't clearly measure in daily operations.

This creates a structural challenge for labs like Anthropic and OpenAI, whose valuations depend partly on continued growth in high-margin API revenue from advanced users. If those users migrate to cheaper open-source alternatives that are only months behind in capability, frontier labs may need to prove concrete ROI — not just benchmark superiority — to justify premium pricing going forward.