OpenAI's GPT-5.6 Sol Fuels Revenue Surge, Overtakes Anthropic

What happened

OpenAI says its GPT-5.6 Sol model, launched in early July, has driven a sharp jump in revenue, with the company reporting quarterly revenue up 35 percent since the release and enterprise revenue growing more than 50 percent over the same period. Data from spend-management platform Ramp indicates OpenAI has overtaken Anthropic in business API spending for the first time in recent months, reversing a trend that had seen Anthropic pull ahead of OpenAI in quarterly revenue.

The shift suggests GPT-5.6 Sol has been particularly effective at winning back enterprise customers and developers who had been routing workloads to competing models, though neither OpenAI nor Ramp has disclosed absolute revenue or spend figures behind the percentage changes.

Why it matters

This is fundamentally a story about the competitive dynamics of the large language model market and how quickly momentum can shift between providers. Enterprise buyers are increasingly multi-homing across model providers and moving spend fluidly based on capability, price and reliability — a single model release can visibly move market share within a single quarter, which is a new kind of volatility for infrastructure-level technology.

For organisations building on top of these models, the lesson is less about which vendor is "winning" this quarter and more about the fact that the underlying assumption of vendor lock-in in AI is weakening. Procurement and platform teams should expect this leapfrogging to continue, and should architect systems that can swap or blend model providers rather than betting entirely on one.

By the numbers

• 35% — OpenAI's reported quarterly revenue increase since GPT-5.6 Sol launched in early July
• 50%+ — growth in OpenAI's enterprise revenue over the same period
• First time — OpenAI has overtaken Anthropic in business API spending, according to Ramp data, since Anthropic previously led on quarterly revenue

The Renascence take

The headline numbers are a useful signal of momentum, but the more interesting story is behavioral: enterprise AI spend is now behaving like a fast-moving consumer market rather than a sticky infrastructure decision, and that has real implications for how experience and technology leaders should plan.

Most coverage will read this as a scoreboard update — OpenAI ahead, Anthropic behind, next quarter reversed. The behavioral reality is that enterprise buyers are optimising for whichever model performs best on their specific workload this month, and switching costs are lower than most vendors assume. Customer-obsessed operators should treat model choice as a live, monitored decision rather than a one-off procurement event, and should design their AI-dependent services so the underlying model can be swapped with minimal disruption to the end-user experience. The organisations that win here won't be the ones that pick correctly once — they'll be the ones that never have to bet the whole experience on a single vendor's roadmap.