The era of corporate blank checks for frontier AI models is officially over, meeting its first cold reality check in the spreadsheets of CFOs. Data from the financial platform Ramp suggests that Anthropic’s Fable 5—widely hailed as the most capable intelligence on the market—is effectively DOA in the enterprise sector. In its debut month, Fable 5 managed to capture a dismal six percent of tokens purchased via Anthropic’s API. Even when measured by total spending, it accounts for a mere 11.4 percent. This isn't just a slow start; it's a clear signal that the corporate appetite for marginal reasoning gains at exorbitant costs has hit a hard ceiling.
The Marginal Utility Trap
In the cold light of the quarterly report, business logic is finally steamrolling technical benchmarks. Anthropic has priced Fable 5 at $10 per million input tokens and $50 per million output tokens—essentially double the cost of OpenAI’s GPT-5.6 Sol. As Ara Kharazian, an economist at Ramp, points out, this premium creates a price ceiling that few organizations are willing to break. For the vast majority of enterprise tasks, the performance edge Fable 5 offers is either irrelevant or so subtle that it vanishes in daily operations.
The performance edge Fable 5 offers may not matter for many use cases, or it's barely measurable in daily work.
When the return on investment remains an abstract promise rather than a calculable figure, companies retreat to safety. We are seeing a massive migration toward 'good enough' models that offer predictable utility over high-cost experimental reasoning. The 'genius' model is becoming a luxury item that businesses simply don't need to balance their books.
OpenAI Wins on Cost-Efficiency
The market penetration of OpenAI’s GPT-5.6 Sol proves that efficiency beats raw power every time. GPT-5.6 Sol currently commands 25 percent of tokens and 23 percent of spending within the OpenAI ecosystem, crushing Fable 5’s adoption rates. According to the Ramp report, Fable 5 generated only 75 percent of the revenue seen by GPT-5.6 Sol, despite Anthropic's significantly higher pricing. This isn't just a win for Sam Altman; it’s a validation of a strategy that prioritizes the price-to-performance ratio. Furthermore, advanced users are increasingly defecting to open-source alternatives that trail frontier models by only a few months, rendering the premium price tag of Fable 5 even more indefensible.
A Crisis for Scaling Laws
This stagnation is a direct threat to the core investment thesis of the major AI labs. While July saw a modest 1.1 percentage point uptick in companies paying for Anthropic, OpenAI managed a pathetic 0.23 percentage points. Meanwhile, xAI has been posting its fastest growth since July 2025, suggesting that the incumbents are losing their grip. The industry is hitting a wall where simply adding parameters and hiking prices no longer moves the needle on market share. For labs like Anthropic and OpenAI, the mission must shift immediately from scaling raw intelligence to radical cost reduction.
Anthropic built a magnificent machine, perhaps the most capable ever seen, only to realize the market is more interested in the price tag than the IQ. The original hype promised that exponential power would lead to exponential revenue, but the data tells a different story. In the real world, businesses prefer a reliable, affordable calculator over a digital savant they can't justify on a balance sheet.