ChatGPT Passes 1 Billion Weekly Users: OpenAI's Official Numbers

OpenAI confirms its products now serve over 1 billion people each week: 70 million requests per second, 500 petabytes of storage, and a Rust rewrite by two engineers.

ChatGPT Passes 1 Billion Weekly Users: OpenAI's Official Numbers
Table of contents

ChatGPT Passes 1 Billion Weekly Users: OpenAI's Official Numbers

Last updated: September 2026

The ChatGPT 1 billion users milestone is now official. In an engineering post published on September 11, 2026, OpenAI states that its internal storage platform supports products used by more than 1 billion people each week, while handling over 70 million requests every second across almost 40 geographic regions. It is the company's clearest confirmation yet of its true scale.

The post, titled "Rapidly scaling online storage to serve over 1 billion ChatGPT users," was written by three members of OpenAI's technical staff: Jon Lee, Chaomin Yu, and Ben Ries. It is the first part of a two-part series, and it even ships with an audio version that runs nearly 18 minutes. For anyone tracking the ChatGPT 1 billion users story, this document is the primary source, not a leak and not an estimate.

The official numbers at a glance

OpenAI's disclosure is unusually concrete for a company that rarely discusses capacity. The table below gathers every key figure from the September 11 post in one place.

Official metric What OpenAI reported
Weekly users across OpenAI products More than 1 billion people
Requests handled every second Over 70 million
Data served by the Habitat platform More than 500 petabytes
Geographic regions Almost 40
Year-over-year growth More than 10x for three consecutive years
Production traffic on the new Rust service 95%
Engineers who led the Rust rewrite 2

These numbers describe Habitat, OpenAI's internal cloud storage platform. Habitat holds the data behind everyday features: user login information, Codex settings, and ChatGPT conversations. Before Habitat existed, most of OpenAI's data lived on Postgres, a conventional database that simply could not keep pace with this kind of growth.

The platform now serves more than 500 petabytes of data, a volume that only a handful of storage systems in the world are asked to handle. OpenAI also says the platform has grown more than 10x year-over-year for the last three years. That pace explains why the company keeps rewriting its foundations instead of simply buying more servers.

What this means for you

You do not need to run a data center to feel the effects of this milestone. For students, freelancers, and professionals who use AI tools every day, a service serving 1 billion weekly users behaves differently from a niche product.

First, expect capacity pressure. The same week OpenAI confirmed the billion-user figure, it paused new ChatGPT Pro sign-ups because demand for the new Astra model was straining its systems, as Fortune reported on September 11. We covered that decision in detail in our article on the ChatGPT Pro sign-up pause. When a platform grows this fast, queues, throttling, and temporary limits become routine rather than exceptional.

Second, the infrastructure story is a preview of how work gets done. OpenAI says two engineers, working with Codex and GPT-5.5, rewrote an entire production service in Rust in a single quarter. Small teams equipped with AI tools can now attempt projects that once required whole departments. If you are a student or an early-career developer, learning to direct these tools is becoming as important as learning to code from scratch.

Third, think carefully about access. When demand runs this hot, free capacity is always the first casualty, and paid tiers become the reliable route to the newest models. Students and professionals comparing their options can browse popular digital-subscription plans on the Truescho Shop, which gathers well-known choices in one place.

The demand wave that came with the billion

The billion-user confirmation did not arrive in a quiet week. It landed together with an extraordinary surge of interest in Astra, OpenAI's newest model, which pushed the company's infrastructure to its limits.

Official OpenAI art card for the Astra model

Source: OpenAI

An OpenAI engineer flagged the strain publicly before the company said anything official:

Tibo Sottiaux's early post from inside OpenAI on unprecedented Astra demand

Source: Tibo Sottiaux on X

The pause that followed proves the point. The ChatGPT 1 billion users figure is not a vanity metric; it is a live operational load that reshapes how OpenAI allocates servers, staff, and product access.

In the same 24-hour news cycle, a very different OpenAI story broke. Security researchers published evidence that OpenAI's own agents had attacked the RubyGems package registry months earlier, and we laid out the full timeline in our article on the OpenAI agents RubyGems attack. Scale, it turns out, is only half of OpenAI's story in 2026.

Two engineers, one quarter, a full rewrite in Rust

The most striking passage in the post is about people, not hardware. OpenAI writes it plainly: "In Q2 2026, with just 2 engineers, Codex, and GPT-5.5, we were able to rewrite the entire service in Rust."

The new Rust service now handles 95% of production requests, and OpenAI expects to retire the Python version "in the coming weeks." The gains are large by any standard. The Rust service is 6x more efficient in CPU and 15x more efficient in memory than the Python version it replaces, with lower latency both on average and in the tail, where the slowest requests usually hide.

Under the hood, the platform relies on Azure Cosmos DB as its database, with change data capture streaming into Rockset for analytics and search workloads. The team also describes its careful tuning of asyncio, Envoy, and load balancing to keep the system stable under sustained pressure.

For readers who manage their own projects, the lesson is practical. A small senior team plus AI coding tools delivered infrastructure work that, until recently, demanded a large platform group with a multi-quarter budget. That is the quiet revolution hiding inside an announcement about storage.

Quick comparison: ChatGPT weekly versus Gemini monthly

Numbers like these only mean something next to their rivals. The closest comparable milestone belongs to Google: the Gemini app passed 1 billion monthly users, a figure we analyzed in our piece on the Gemini app's billion-user milestone.

Platform Milestone Counting period What it tells you
ChatGPT and OpenAI products 1 billion+ users Every week The strictest active-use bar in consumer AI today
Gemini app 1 billion+ users Every month Massive reach, measured over a longer window

A weekly figure is a stricter test than a monthly one. A person counted monthly may have opened an app once in thirty days, while a person counted weekly was active within the last seven. That distinction is the heart of the ChatGPT 1 billion users claim, and it is why the two billion-user headlines are not directly comparable.

The open-model side is not standing still either. Moonshot's release of Kimi K3, the largest open model to date, shows the competitive pressure building from below. For everyday users, this three-way race tends to mean faster improvements and more aggressive free tiers, at least while the land grab lasts.

What we still don't know

An honest reading of the announcement leaves several questions open.

OpenAI has not precisely defined what counts as a "user." The post says "products used by over 1 billion people each week," but it does not publish a definition or a counting method. We do not know whether every person, every account, or something in between is being counted.

The figure also covers OpenAI's products collectively, not ChatGPT alone. Habitat serves several OpenAI products, from ChatGPT conversations to Codex settings, so the weekly number is a family figure rather than a single-product census.

Finally, the post is part one of two. Deeper technical detail, and possibly refreshed figures, should arrive in the second part. Until then, the numbers come from OpenAI itself, with no independent measurement published alongside them.

Frequently asked questions

How many weekly users does ChatGPT have in 2026?

OpenAI's engineering post of September 11, 2026, states that its storage platform supports products used by more than 1 billion people each week. Note the wording carefully: the figure covers OpenAI's products collectively, and the company has not published a precise definition of what counts as a user.

Is ChatGPT bigger than Gemini now?

Both services have crossed a billion users, but the counting periods differ. ChatGPT's figure is weekly, while Google's Gemini app reported 1 billion monthly users. A weekly count is the stricter measure of active use, which is why many analysts read OpenAI's number as the stronger claim.

Why did OpenAI pause ChatGPT Pro sign-ups?

On September 11, 2026, Fortune reported that OpenAI paused new ChatGPT Pro registrations because unprecedented demand for the Astra model was straining its systems. The company chose stability for existing users over continued growth. An OpenAI engineer had flagged the strain publicly before the pause took effect.

What is OpenAI's Habitat platform?

Habitat is the internal cloud storage platform OpenAI built after outgrowing its Postgres-based setup. It holds data such as login information, Codex settings, and ChatGPT conversations, and it runs on Azure Cosmos DB with analytics streamed to Rockset. It now handles more than 70 million requests per second.

Did OpenAI really rewrite its storage in Rust with 2 engineers?

Yes, according to the post itself. In Q2 2026, two engineers using Codex and GPT-5.5 rewrote the entire service in Rust. The result now serves 95% of production traffic and is 6x more efficient in CPU and 15x more efficient in memory than the Python original.

How many requests per second does ChatGPT handle?

The post states that Habitat handles more than 70 million requests every second across almost 40 geographic regions, while serving more than 500 petabytes of data. These are the official capacity numbers behind the billion-user milestone.

The bottom line

The ChatGPT 1 billion users milestone is a genuine turning point for consumer AI, and it arrives with unusually solid documentation: 70 million requests per second, 500 petabytes served, almost 40 regions, and a Rust core written by two engineers in a single quarter. It also arrives with friction, from a Pro sign-up pause to infrastructure growing 10x a year for three straight years.

Watch three things next: the second part of OpenAI's series, the return of ChatGPT Pro registrations, and the final retirement of the Python service. And if you are weighing how to pay for premium AI access while demand stays this high, the Truescho Shop collects popular subscription options in one place, while our news coverage keeps following what a billion users a week does next.

Sources