More than half of ChatGPT’s active users now primarily use a language other than English, as reported in consumer usage data released by OpenAI. The most common languages other than English are Spanish, Portuguese, and Arabic.
Figures include individual ChatGPT plans, such as Free, Go, Plus, and Pro, but do not include Codex, Enterprise, or Education products. This means that they are not ideal for use in workplaces, technical fields, or classrooms. Among this group of consumers, OpenAI says the use of non-English languages has increased in line with the growth of the platform since its launch in 2022.
What the Data Shows
OpenAI’s Signals program tracks how consumers use their programs over time. Measured against a July 2023 baseline, ChatGPT’s weekly active users grew across all continents, with the fastest relative growth in Africa and Asia.
OpenAI reports that countries with a low HDI had the fastest relative growth during that period, and notes that it continues to provide low-cost access through Free and Go programs.
In that non-English profession, the minority languages are growing very fast. In June 2026, Uzbek, Kazakh, and Burmese saw the largest increase in active users among languages with more than 1 million users.
A different analysis shows that after six months on ChatGPT, people send about 50% more messages every day and try twice as many unique activities. This data, taken from a small sample of 0.1% of accounts created between October 2025 and early May 2026, provides a snapshot of recent user activity through the end of that period. Remember, though, it doesn’t show the entire history of the field.
OpenAI estimates that people with predominantly female names now lead usage among unsplit names, after reaching parity last year, by increasing demographics using name-gender matching instead of user data.
Why This Matters
ChatGPT’s user base has changed from a predominantly English-speaking population to a more geographically diverse, less English-speaking group than the original application.
That is always wrong with the way the retrieval layer behaves. I put together a report earlier this year showing that ChatGPT Search often runs background queries in English, even if the actual information is not in English, meaning that a Spanish or Arabic query can still draw on English sources before reaching the answer.
This is the second time in less than a year that OpenAI has released consumer spending data for policy-facing research. When the company published a major retrospective of consumer spending with Harvard last September, the takeaway was a narrowing of the gender gap and rapid growth in low-income countries. This release expands that image.
Looking Forward
Signals are a continuous release, so it’s interesting to see how the majority of non-English people appear here.
The numbers are even higher for businesses that focus on an English-speaking audience. If the majority of people within ChatGPT speak another language, an English-only audience is no longer an audience.
