OpenAI says its models now reach 1 billion people every week

OpenAI Says Its Models Now Reach 1 Billion People Every Week

OpenAI says its AI models now reach more than 1 billion people every week.

The milestone is real, and it is fast. OpenAI announced on July 31, 2026, that its models surpass 1 billion weekly active users, less than four years after ChatGPT launched in November 2022 1. The company disclosed the figure in a blog post, reporting that its models now reach "more than one billion active users and more than two million businesses" 2. For comparison, Facebook took six years after launch to reach 1 billion users. Google Maps, TikTok, Instagram, and YouTube each needed between five and eight years 3.

But OpenAI was actually late to its own target. The company had projected crossing 1 billion weekly users about seven months earlier, according to reporting by The Information 3.

The 1 billion figure also covers more than ChatGPT alone. According to the company, it includes Codex, its model dedicated to AI-assisted programming, and ChatGPT Work, a platform that allows users to access external applications 2.

Then came the price cuts. Luna, released in early July as part of a three-model GPT-5.6 family, already cost roughly one-fifth as much as Sol, OpenAI's highest-performing variant. The 80 percent reduction now places Luna among the least expensive advanced models available, including those developed in China 2.

OpenAI framed both moves under a banner it calls "building abundant intelligence," writing in its blog post: "Our goal is not simply more compute, bigger models, or lower token prices. It is more useful intelligence within reach" 1.

Notice the contradiction. "Lower token prices" is listed as something the company says it is not optimizing for. The announcement of the deepest price cut in the model's brief history lands on the same page.

The real story is about leverage. When you reach 1 billion people every week, you can sell intelligence at a fraction of what it costs to produce. You can absorb losses that would break a smaller rival. And you can lock in developers who, once they build applications on your cheapest tier, face real switching costs to leave.

That is why the milestone and the price cut arrived together. OpenAI is converting a distribution advantage into a pricing weapon.

For competitors, the pressure is severe. Anthropic, which focuses on enterprise clients, recently surpassed OpenAI in revenue, according to a report by The Information 3. OpenAI's chief financial officer Sarah Friar said business clients now account for 40 percent of revenue, with that share expected to reach 50 percent by year's end 3. The company projects $25 billion in revenue this year, against a projected $25 billion in cash burn 3.

Burning $25 billion to sustain a billion-user platform is not a vulnerability. It is a moat. OpenAI is spending at a scale few competitors can match, and cutting prices to a level few competitors can survive. Meanwhile, the cost of running AI has climbed sharply as agents, which perform tasks autonomously and evaluate their own results, drive sustained demand for computing power 2. Developers and enterprises feeling that cost pressure are already turning to cheaper alternatives, including open-weight models from China 2.

For builders, the signal is that model quality across the frontier is converging. The decision about which model to build on is becoming less about benchmark scores and more about cost, lock-in, and distribution reach. For investors, it is that the frontier-model market is commoditizing faster than most pricing models assumed. For enterprises, it is that the cheapest intelligence will come from whichever company can afford to lose the most money delivering it.

OpenAI's message to competitors is buried in the numbers, not the blog post. If you cannot match a billion users and an 80 percent price cut at the same time, the commoditization war is already over. You just have not been told yet.