Yesterday evening, Sam Altman shared an image of the Death Star on X. There was no caption on the picture, which showed the world-destroying Star Wars space station rising over an Earth-like planet, but his audience understood the context. In less than 24 hours, OpenAI would release an AI model intended to wipe out all the rest.
That model, GPT-5, indeed launched earlier today with all the requisite fanfare. In an announcement video, Altman said that the product will serve as a “legitimate Ph.D.-level expert in anything—any area you need, on demand—that can help you with whatever your goals are.” He added that, “anyone, pretty soon, will be able to do more than anyone in history could.” In more concrete terms, GPT-5 is an upgrade to the ChatGPT interface you’re likely already familiar with: a model that’s now a bit better at writing, coding, math and science problems, and the like.
Of course, Altman has a penchant for hyperbole, and OpenAI—like the rest of the AI industry—likes to tout each new model as the best ever. But this particular release feels notable for a few reasons. First, it has been a long wait since the release of GPT-4 in March 2023, just a few months after ChatGPT’s debut in November 2022. And second, in that time, OpenAI has become a bona fide tech empire: As of this week, OpenAI now provides enterprise ChatGPT accounts to federal agencies at essentially no cost; its products are also used by nearly every Fortune 500 company; and today Altman announced that roughly 700 million people worldwide use ChatGPT every week. In terms of sheer reach, this is the company’s most consequential product announcement, ever.
As OpenAI has ascended to the scale of a typical tech giant—as of this week, it is reportedly in talks for a $500 billion valuation—the firm has also started to act like its corporate rivals. To attract new users and customers (and keep existing ones from turning to other AI products), OpenAI has doubled down on institutional partnerships and polishing its product lineup. Sure, the company still pushes the limits of AI capabilities—but its products are what keep most consumers and businesses coming back for more. For instance, OpenAI has partnered with Bain & Company, Mattel, Moderna, Los Alamos National Laboratory, and Harvard. It has brought on Jony Ive, the designer of the iPhone, to spearhead the creation of physical OpenAI devices. (The Atlantic and OpenAI have a corporate partnership.)
GPT-5 achieves state-of-the-art performance on a number of AI benchmarks, according to OpenAI’s internal tests, but it is far from a clean sweep: There are a few tests on which competing products such as Google Gemini, Anthropic’s Claude, or xAI’s Grok outperform, or are just barely below the level of, OpenAI’s new top model. The GPT-5 announcement video and launch page also contained a number of errors—incorrect labels, numbers and colors that made no sense, and missing entries on charts—that made the program’s precise abilities, and the trustworthiness of OpenAI’s reporting, hard to discern (and led some observers to joke that perhaps GPT-5 itself had made, or hallucinated, the graphics). Yet that may not matter. OpenAI’s animating theme for GPT-5 is user experience, not “intelligence”: Its new model is intuitive to use, fast, and efficient; adapts to human preferences and intentions; and easily personalizable. Before it is more intelligent, GPT-5 is more usable—and more likely to attract and retain users. “The important point is this,” Altman said, pinching a thumb and index finger together for emphasis: “We think you will love using GPT-5 much more than any previous AI.”
In some sense, OpenAI is learning from its greatest success. ChatGPT took off because it effectively redesigned an existing product: GPT-3.5, ChatGPT’s original underlying model, was months old by the time the chatbot came out, but it was relatively obscure. Placing essentially the same program within a conversational interface, however, made the model easy to use and obsess over. GPT-4 would eventually provide a new engine—smarter and more capable—but this was almost beside the point; to most people, the product was already firmly established as ChatGPT. And, like the original ChatGPT, GPT-5 is free, although nonpaying users have a limit on their usage of this most advanced model—giving everyone a small taste of OpenAI’s ecosystem to open up the possibility they will want, and pay, for more.
During the ensuing two-plus years of the AI race, OpenAI has kept up by releasing a slew of more minor models and new features. When Google released a version of Gemini that was extremely fast and cheap, OpenAI did the same; when DeepSeek launched a free and advanced model that could “reason” through complex questions, OpenAI publicly released a still more powerful reasoning system of its own; as Anthropic’s Claude Code seemed to corner the AI-coding market, OpenAI came out with the Codex tool for software engineers. The empire’s ambitions had no limits.
But these products were accompanied by a labyrinth of names and uses: GPT-4o and GPT-4o mini and GPT-4.1; o1-mini and o1-pro; o3 and o3-pro and o4-mini; and so on. This was a matter not only of poor branding but of poor design. Despite the numbers, for some uses o3 is better than o4. Users frequently complain that they don’t know how to select from OpenAI’s models. “We are near the end of this current problem,” Altman said on OpenAI’s podcast in June. “I am excited to just get to GPT-5 and GPT-6, and I think that’ll be easier for people to use.”
Now OpenAI has arrived at GPT-5, and indeed, the model might be best understood as providing easier and frictionless use—as an amalgam of all of OpenAI’s disparate, discrete advances from the previous two-plus years. GPT-5 “eliminates this choice” among models and their specialties, Mark Chen, OpenAI’s chief research officer, said in today’s announcement, and that may be the new model’s core feature. GPT-5 modulates its approach to your query, using more or less “reasoning” power—doing the equivalent of selecting among the GPT-4os and o3s and o4s—depending on what is asked of it. OpenAI is now retiring a large number of its previous, major models.
Alongside GPT-5, OpenAI also announced a number of other additions to the ChatGPT experience to “make ChatGPT more personalized,” Chen said, “so it’s more like your AI.” These new features are customizable color schemes, personalities (“cynic,” “robot,” “listener,” “nerd”), and access to Gmail and Google Calendar—all building on top of the recently added “Memories” feature, through which ChatGPT can pull information from previous chats. These add-ons have little to do with the bot’s engine—how “intelligent” or “capable” it is—but they will make ChatGPT more customizable, more useful, and perhaps more fun. Businesses can integrate their data, as well. Just as the years of photos and notes on your iPhone make it undesirable to switch to a Google Pixel, or years of using Google Drive make it hard to migrate to Microsoft OneDrive, if ChatGPT morphs from a vanilla bot into your AI or your company’s AI, leaving for Gemini or Claude becomes not just burdensome, but a downgrade.
At this stage of the AI boom, when every major chatbot is legitimately helpful in numerous ways, benchmarks, science, and rigor feel almost insignificant. What matters is how the chatbot feels—and, in the case of the Google integrations, that it can span your entire digital life. Before OpenAI builds artificial general intelligence—a model that can do basically any knowledge work as well as a human, and the first step, in the company’s narrative, toward overhauling the economy and curing all disease—it is aiming to build an artificial general assistant. This is a model that aims to do everything, fit for a company that wants to be everywhere.
