To answer the first question, the researchers created a case study they called the GPT-5 package, which they say included all OpenAI offerings available during the lifetime of GPT-5 as a flagship model, including GPT-5 and GPT-5.1, GPT-4o, ChatGPT, and the API, and estimated the package’s revenue and operating costs. All the numbers compiled were based on information sources that included claims by OpenAI and its employees, and media reports, primarily The Information, CNBC, and the Wall Street Journal.
The revenue estimate is “relatively straightforward,” they said. Since the package included all OpenAI models, this was the company’s total revenue for the lifetime of GPT-5 from August to December of last year: $6.1 billion.
And they pointed out that “at first glance, $6.1 billion sounds healthy until you compare it to the cost of running the GPT-5 package.” Those costs come from four main sources, the report says, the first of which is put at $3.2 billion. This figure is based on public estimates of OpenAI’s total computing expenditure in 2025 and assumes that the allocation of computing during GPT-5’s tenure was proportional to the fraction of annual revenue generated during that period.