Operator
Product information, use cases, and access for Operator.
Pricing information
No verified public pricing is available yet.
DevPrice organizes public information and does not sell Operator subscriptions. Prices and availability are determined by Operator.
What is Operator?
Operator is an AI agent from OpenAI designed to carry out browser-based tasks from a user’s description. It observes what is shown on a web page and interacts with the interface through virtual mouse and keyboard actions. This makes it suitable for online work such as making reservations, shopping, and completing forms when the task spans several pages or steps.
The agent can reason about the current state of a task and adjust its next action when the page does not behave exactly as expected. It can also work on multiple tasks and use stated preferences to avoid repeating the same instructions. When a task reaches a sensitive point such as entering login or payment information, Operator can stop and hand control back to the user so the user remains responsible for the consequential action.
Key features of Operator
Browser interface control
Operator reads the visible browser interface from screenshots, identifies elements such as buttons, menus, and text fields, and uses virtual mouse and keyboard actions to click, scroll, and type.
Online task execution
A user can describe an objective and Operator can carry it through a sequence of browser steps for activities such as reservations, online shopping, or form completion, stopping when the task is done or human input is needed.
Reasoning and adjustment
When a page changes or an action does not produce the expected result, Operator reassesses the new visual state, adjusts its next move, and returns control to the user if it cannot proceed.
Parallel work and handoff
Operator can run more than one online task and retain user-provided preferences for recurring work. At sensitive stages involving login or payment, it pauses the workflow and lets the user take over or confirm the next action.