Ddos 
OpenAI has officially unveiled the ChatGPT Agent—a groundbreaking capability endowed with general operational intelligence that transforms AI from a passive responder into an active executor of complex user tasks.
This new feature is currently being rolled out to ChatGPT Pro, Plus, and Team subscribers, with potential plans to make a limited version available to free-tier users in the future.
The ChatGPT Agent introduces powerful functionalities such as automatic calendar navigation, presentation generation, slide drafting, and even code execution within an integrated terminal. It seamlessly incorporates several of OpenAI’s recent technological breakthroughs, including the Operator (capable of clicking and interacting with websites) and the Deep Research module, designed to aggregate information and produce comprehensive reports.
Activating the feature is remarkably straightforward. Users simply enable Agent mode via the ChatGPT tool menu to access the Agent interface. From there, natural language prompts can initiate intricate workflows—such as planning and booking a U.S. vacation, sourcing ingredients for a Japanese breakfast, or analyzing three market competitors and generating a polished presentation. The system also supports various Connectors, enabling integrations with services like Gmail and GitHub, or leveraging APIs to interact with external applications, thereby orchestrating more sophisticated task chains.
As part of this upgrade, OpenAI has announced that the Operator feature will be retired in 30 days. However, the Deep Research module will continue to be offered as a standalone tool for use cases requiring in-depth analysis.
According to OpenAI’s published test results, the ChatGPT Agent has delivered impressive performance across multiple AI benchmarks. On “Humanity’s Last Exam (pass@1),” it achieved a 41.6% success rate—roughly double that of o3 and o4-mini models. In the “FrontierMath” assessment, it scored 27.4%, significantly surpassing o4-mini’s 6.3%. These results underscore that the Agent is not merely a language model, but a versatile AI system capable of reasoning and execution.
As agent-based AI tools begin to penetrate mainstream markets, OpenAI emphasizes the paramount importance of safety. For operations involving travel bookings, form submissions, or interactions with personally identifiable information, the ChatGPT Agent will always request explicit user authorization beforehand.
Additionally, OpenAI introduces a “supervised mode,” requiring user approval for each action when handling high-risk tasks. For sensitive matters such as financial transactions or legal advice, the Agent is designed to proactively decline execution.
While several AI agents have emerged in the market, most remain limited when handling complex operations. OpenAI asserts that, powered by its most advanced foundational models and a robust integration framework, the ChatGPT Agent stands as the most competitive and capable general-purpose AI agent available today—elevating ChatGPT from a responsive tool to a truly executional AI platform.
Donate