OpenAI today launched ChatGPT agent, a new agentic model that is able to think proactively and complete computer-based tasks on the user's behalf.
The ChatGPT agent is in the same family as o3. It combines several existing ChatGPT features, and it can do things like research and generate reports, execute code using Terminal, generate slides and spreadsheets, and connect to external data sources and applications.
OpenAI gives several examples of how ChatGPT agent can be used:
- Look at my calendar and brief me on upcoming client meetings based on recent news.
- Plan and buy ingredients to make Japanese breakfast for four.
- Analyze three competitors and create a slide deck.
The ChatGPT agent uses its own virtual computer, and it will navigate websites, filter results, prompt users to log into websites when needed, and deliver summaries of its findings. It is designed to seek permission before taking any "actions of consequence," and OpenAI says that users can interrupt tasks to add extra instructions, and stop tasks at any point.
ChatGPT agent is rolling out starting today for Pro, Plus, and Team users. Just select "agent mode" from the dropdown menu in the composer during a conversation. ChatGPT users are able to transition between conversations and action requests within the same chat.
Pro users will get access by the end of today, while Plus and Team users will get access over the next few days. OpenAI plans to add the functionality for Enterprise and Education users in the coming weeks. Pro users have access to 400 messages per month, and other paid users will get 40 messages monthly with additional usage available through flexible credit-based options.
