Corporate AI Spending: OpenAI's GPT-5.6 Sol Outpaces Anthropic's Fable 5
New corporate spending data reveals that while Anthropic leads in overall AI adoption, OpenAI's GPT-5.6 Sol model generated more enterprise revenue in the U.S. in July than Anthropic's Fable 5.

OpenAI's GPT-5.6 Sol artificial intelligence model generated more revenue from U.S. businesses in July than Anthropic's Fable 5, the company's most advanced offering, according to new data from fintech firm Ramp. The AI Index report, which tracks spending across over 70,000 U.S. businesses, indicates a shift in enterprise AI spending patterns.
While Anthropic continues to lead in overall enterprise adoption of AI, the revenue figures suggest a stronger performance from OpenAI's model within the high-end market. Businesses spent approximately 75% of what they spent on OpenAI's GPT-5.6 Sol on Anthropic's Fable 5 in July. This could signal either a maturing market where companies are scrutinizing costs for top-tier AI or a growing preference for OpenAI's solutions at the higher end.
Ramp's analysis, derived from anonymized billing data, provides insight into AI service adoption. A weaker demand for the most expensive AI models might indicate that enterprise clients are beginning to curb their AI expenditures. This trend could be a concern for Anthropic, which is reportedly banking on users being willing to pay a premium for its most advanced AI technologies ahead of a potential IPO.
Despite these specific revenue figures, Anthropic still shows strong growth and leads in overall enterprise adoption. In July, 43.5% of U.S. businesses used Anthropic AI products, compared to 39.7% for OpenAI. Google and xAI followed with 6.2% and 4% adoption rates, respectively.
The report also highlights the growing importance of AI agents, tools capable of automating tasks and potentially snowballing usage costs. Ramp notes that the significant growth in AI spending is largely driven by these agents, attributing it to their scalability, which surpasses that of traditional software.