OpenAI Says It Has 1 Billion Active Users and More Than 2 Million Business Customers

OpenAI’s 1 billion figure is missing a denominator that matters for commercial analysis. The company has not explained whether active accounts correspond to unique people, nor has it separated free users, individual subscribers, and enterprise seats. If paid seats, API volume, and revenue do not rise with the active-user count over the next three to six months, the milestone will demonstrate reach without showing that revenue per user has increased.

OpenAI CFO Sarah Friar wrote on July 31, 2026 that the company’s models now reach more than 1 billion active users and more than 2 million businesses. The Verge described the first figure as weekly active users, while PYMNTS used “active users” without specifying a period. The material available from OpenAI does not provide account deduplication, regional distribution, the paid share, or a measurement window. For that reason, the 1 billion number cannot be compared directly with another platform’s monthly-active-user figure.

Message volume rises after six months

Friar also supplied a measure of usage depth. After six months with the technology, individual users send 50% more messages per day and use ChatGPT for twice as many kinds of work. Businesses often begin with one team or workflow and then expand adoption across operations. These observations come from OpenAI, which did not publish the sample size, explain whether churned accounts were included, or isolate the effect of product changes during the same period. The data describes how retained users deepen their use; it does not establish that every new account develops the same habit.

The claim of more than 2 million businesses also needs a distinction between registration and paid deployment. OpenAI did not report average seats per business, renewal rates, API spending, or the percentage of workflows in production. A larger customer count gives an enterprise software vendor more sales entry points. Durable departmental use and payment would have to appear in revenue per customer and retention before the count demonstrates commercial depth.

Luna falls 80%, Terra falls 20%

One day before the user announcement, OpenAI changed the price or performance of three GPT-5.6 models. GPT-5.6 Luna became 80% cheaper, GPT-5.6 Terra became 20% cheaper, and GPT-5.6 Sol became faster in the API while keeping the same price. Friar said these changes let customers balance intelligence, speed, reliability, and cost. Lower prices can make high-volume or batch jobs economical enough to test, but the cited sources do not disclose unit inference cost, gross margin after the cuts, or demand elasticity.

The Verge independently confirmed that OpenAI announced the user milestone and the two price reductions. PYMNTS reported the business count and the usage behavior after six months. The central figures in both reports still originate with the company and have not been independently audited. Over the next three to six months, useful indicators are whether the base grows beyond more than 2 million businesses, whether messages per active user remain elevated, and whether API volume offsets the 80% price reduction. If OpenAI updates only the headline user count while withholding paid conversion and retention, outsiders still cannot translate scale into sustainable revenue.

Sources:

• OpenAI: Building abundant intelligence
• The Verge: OpenAI says its models now reach more than 1 billion weekly active users
• PYMNTS: OpenAI Reaches 1 Billion Active Users as AI Becomes Daily Habit