OpenAI’s next move in its battle against Google is an AI-powered web browser. The tool, dubbed ChatGPT Atlas, is out today. The company announced it in a livestream after teasing it earlier Tuesday via a mysterious video of browser tabs on a white screen.

ChatGPT Atlas is available “globally” on macOS starting today, while access for Windows, iOS, and Android is “coming soon,” per the company. But its “agent mode” is only available to ChatGPT Plus and Pro users for now, OpenAI CEO Sam Altman said on the livestream.

“The way that we hope people will use the internet in the future… the chat experience in a web browser can be a great analog,” Altman said.

Besides Altman, the livestream featured OpenAI employees Will Ellsworth, who works on post-training research; Adam Fry, the product lead for ChatGPT Search; Ben Goodger, a staff member who in previous roles helped develop Google Chrome and Mozilla Firefox; Ryan O’Rouke, an interface designer; Justin Rushing, who previously worked at Apple; and Pranav Vishnu.

Fry said one of the browser’s best features is memory — making the browser “more personalized and more helpful to you,” as well as an agent mode, meaning that “in Atlas, ChatGPT can now take actions for you… It can help you book reservations or flights or even just edit a document that you’re working on.” Users can see and manage the browser’s “memories” in settings, employees said, as well as open incognito windows.

The browser’s agent mode clearly builds on OpenAI’s past forays into agentic AI, such as its Operator tool, an early version of a tool allowing ChatGPT to use a computer on a user’s behalf, and ChatGPT Agent, the next iteration, which was designed to be able to complete more complex tasks, shop, and more (though it wasn’t always successful in that realm).

Whenever you click a link from a search result in Atlas, it’ll by default show a split-screen with the webpage and the ChatGPT transcript, with the goal being to display a “companion” at all times, employees said, though a user can turn off the split-screen if they’d rather. On the livestream, employees also demonstrated the browser’s summarization features for webpages, as well as selecting text from an email and clicking a button to have ChatGPT tidy up the sentence in-line — the latter feature is called “cursor chat.”

“This is just a great browser all-around — it’s smooth, it’s quick, it’s really nice to use,” Altman said.

The AI browser wars have been heating up for a while — OpenAI announced a prototype of its search engine, dubbed SearchGPT, all the way back in July 2024. But in recent months, AI-fueled browsers have felt like the latest frontier in AI hype.

This summer, Perplexity launched its buzzy Comet browser, an AI-powered solution aiming to simplify the way people browse the web and complete tasks. Instead of a laundry list of Google Search results, you get a Perplexity “answer engine,” which offers a few links to relevant websites and generates an answer to your query. It can also scan all your open tabs, summarize videos, declutter your email inbox, and even make purchases on Amazon.

In September, Google announced that it would more deeply embed its Gemini AI assistant into Chrome, and that in the coming months, it plans to allow Gemini in Chrome to be able to do “tedious tasks” on your behalf, such as grocery shop, schedule appointments, book reservations, and more — although Google declined to specify a launch date.