OpenAI has taken a significant leap forward in artificial intelligence with the launch of ChatGPT Agent, a system that can independently complete complex tasks from start to finish using its own virtual computer.
The new agent, announced on July 17, 2025, represents a unified approach that combines three previously separate capabilities: Operator's ability to interact with websites by clicking, scrolling and typing; deep research's skill in synthesizing information from across the web; and ChatGPT's conversational intelligence. This integration addresses limitations of earlier tools that worked well in isolation but couldn't handle end-to-end workflows.
Powered by GPT-4o, OpenAI's flagship multimodal model, ChatGPT Agent can handle sophisticated requests like "analyze three competitors and create a slide deck" or "look at my calendar and brief me on upcoming client meetings based on recent news." The system navigates websites both visually and textually, completes forms, accesses authorized accounts with user permission, executes code, and produces editable documents including spreadsheets and presentations.
In benchmark tests, ChatGPT Agent significantly outperforms previous OpenAI tools. On investment banking analyst modeling tasks, it surpasses both deep research and the o3 model. On the BrowseComp benchmark for locating hard-to-find information, it achieved a new state-of-the-art score of 68.9%, 17.4 percentage points higher than deep research.
While powerful, OpenAI emphasizes that users remain in control. The agent requests permission before taking consequential actions, and users can interrupt, take over the browser, or stop tasks at any point. Starting today, Pro, Plus, and Team users can activate these capabilities through the tools dropdown by selecting 'agent mode' in any conversation.
This launch marks OpenAI's boldest attempt yet to transform ChatGPT from a question-answering tool into an agentic product that can take actions and offload complex tasks for users. While early AI agents have struggled with complex tasks, OpenAI claims ChatGPT Agent is far more capable than previous offerings, with plans for regular improvements to make it increasingly useful over time.