TL;DR: OpenAI just launched ChatGPT Atlas, a desktop web browser with ChatGPT built in. Think split-screen chats beside pages, one-click summaries, and an Agent Mode that can actually complete multi-step tasks on the web. Atlas ships today on macOS with Windows, iOS, and Android "coming soon.”
For years, we copied links into ChatGPT for help. Today, OpenAI flipped that flow: the browser itself now is the assistant. Open a page, and Atlas sits right there—summarizing reports, comparing products, drafting replies—without you ever leaving the tab. In Agent Mode, it can even move through sites step-by-step to get things done (think trip planning or a returns flow), with a running explanation of what it’s doing.
Atlas enters a crowded field Chrome (with Gemini integrations), Microsoft Edge (Copilot), and new AI browsers like Comet. The pitch is different: make the browser your co-pilot by default, not an extension you bolt on. That shift is why many see this as a credible challenge to Chrome’s dominance.
Browsers have long been windows; Atlas turns the window into a collaborator. If OpenAI executes, common "browser chores” (shopping comparisons, travel planning, form wizards) shift from manual clicking to supervised automation—with a readable trail of actions that builds trust. That’s a tangible step toward agentic computing… inside the app people use most.
What is ChatGPT Atlas?
An AI-integrated web browser from OpenAI with ChatGPT built in, including a task-performing Agent Mode.
Is Atlas available now?
Yes—macOS today; Windows, iOS, and Android are coming soon.
How is it different from Chrome with an extension?
Atlas treats ChatGPT as a first-class browsing interface: split-screen chat, page-aware actions, and agentic workflows that can navigate sites for you with transparent step logs.
Does it replace search?
Not entirely. It reframes many "search-and-click” tasks as conversations + actions, while still letting you browse the open web.