{"id":680483,"date":"2025-07-18T15:05:13","date_gmt":"2025-07-18T12:05:13","guid":{"rendered":"https:\/\/buradabiliyorum.com\/en\/openai-launches-a-general-purpose-agent-in-chatgpt-2\/"},"modified":"2025-07-18T15:05:13","modified_gmt":"2025-07-18T12:05:13","slug":"openai-launches-a-general-purpose-agent-in-chatgpt-2","status":"publish","type":"post","link":"https:\/\/buradabiliyorum.com\/en\/openai-launches-a-general-purpose-agent-in-chatgpt-2\/","title":{"rendered":"OpenAI launches a general purpose agent in ChatGPT"},"content":{"rendered":"
\n

OpenAI is launching a new general<\/a> purpose AI agent in ChatGPT, which the company says can complete a wide variety of computer-based tasks on behalf of users. OpenAI says the agent can automatically navigate a user\u2019s calendar, generate editable presentations and slideshows, and run code.<\/p>\n

The tool, called ChatGPT agent, combines several capabilities from OpenAI\u2019s previous agentic tools, including Operator\u2019s ability to click around on websites, as well as Deep Research\u2019s ability to synthesize information from dozens of websites into a concise research report. OpenAI says users will be able to interact with the agent simply by prompting ChatGPT in natural language.<\/p>\n

ChatGPT agent is rolling out on Thursday to subscribers to OpenAI\u2019s Pro, Plus, and Team plans. To activate the tool, users can select \u201cagent mode\u201d in ChatGPT\u2019s dropdown menu of tools.<\/p>\n

\n
\n
\n

ChatGPT can now do work for you using its own computer.<\/p>\n

Introducing ChatGPT agent\u2014a unified agentic system combining Operator\u2019s action-taking remote browser, deep research\u2019s web synthesis, and ChatGPT\u2019s conversational strengths. pic.twitter.com\/7uN2Nc6nBQ<\/a><\/p>\n

\u2014 OpenAI (@OpenAI) July 17, 2025<\/a><\/p><\/blockquote>\n<\/div>\n<\/figure>\n

The launch of ChatGPT agent represents OpenAI\u2019s boldest attempt yet to turn ChatGPT into an agentic product that can take actions and offload tasks for users, rather than just answering questions. In recent years, Silicon Valley companies including OpenAI, Google, and Perplexity have unveiled dozens of AI agents that have promised to do just that. However, these early version AI agents have proven to struggle with complex tasks, and they seem less compelling as products than the ultimate vision tech executives pitch around AI agents.<\/p>\n

That said, OpenAI says ChatGPT agent is far more capable than its previous offerings.<\/p>\n

The company\u2019s new agent can access ChatGPT connectors, allowing users to connect app<\/a>s like Gmail and GitHub so that the agent can find relevant information to your prompts. OpenAI says ChatGPT agent has access to a terminal, and it can use APIs to access certain apps.<\/p>\n

OpenAI suggests that users can tap ChatGPT agent to \u201cplan and buy ingredients to make Japanese breakfast for four,\u201d as well as \u201canalyze three competitors and create a slide deck.\u201d These kinds of capabilities requires ChatGPT agent to parse through websites, plan a course of action, and use tools \u2014 much more complicated tasks than OpenAI has previously tried to tackle with agents.<\/p>\n

\n
\n

Techcrunch event<\/p>\n

\n

\n\t\t\t\t\t\t\t\t\tSan Francisco<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 27-29, 2025<\/span>\n\t\t\t\t\t\t\t<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n

The model underlying ChatGPT agent offers state-of-the-art performance on several benchmarks, according to OpenAI.<\/p>\n

The company says the ChatGPT agent model scores 41.6% on Humanity\u2019s Last Exam (pass@1), a difficult test made up of thousands of questions across more than one hundred subjects. That\u2019s roughly double what OpenAI\u2019s o3 and o4-mini scored on the test.<\/p>\n

On FrontierMath, one of the hardest known math benchmarks, OpenAI says ChatGPT agent scores 27.4% when it has access to tools, such as a terminal for code execution. The previous state-of-the-art score comes from o4-mini, which scored just 6.3%.<\/p>\n

OpenAI notes that it developed ChatGPT agent with safety in mind, largely because the product presents some newfound capabilities that could make it more dangerous in the hands of a bad actor. OpenAI has previously warned that agentic models could present more dangerous capabilities.<\/p>\n

In a safety report<\/a> for ChatGPT agent, OpenAI says it\u2019s designated the model as \u201chigh capability\u201d in biological and chemical weapon domains, which is defined in OpenAI\u2019s Preparedness Framework as a model with the ability to \u201camplify existing pathways to severe harm.\u201d OpenAI notes that it does not have direct evidence of this, but it\u2019s decided to take a precautionary approach and activate new safeguards to mitigate these risks.<\/p>\n

The new safeguards for ChatGPT agent include a monitor that works in real time as users interact with the product. OpenAI says it runs a classifier across every prompt entered into ChatGPT agent, determining whether the request is related to biology. If so, OpenAI runs ChatGPT agent\u2019s response through a second monitor that determines whether the content could be used to evoke a biological threat.<\/p>\n

OpenAI also says it disabled ChatGPT\u2019s memory feature for this agent to prevent misuse. In other parts of ChatGPT, OpenAI\u2019s memory feature allows the chatbot to reference information from previous user chats. However, OpenAI says bad actors could use the feature in ChatGPT agent to exfiltrate sensitive data through prompt injection attacks. The company says it may revisit adding the feature in the future, however.<\/p>\n

While ChatGPT agent sounds impressive, it remains to be seen how capable it truly is in the real world. Until now, agent technology<\/a> has proven relatively brittle when interacting with the real world. That said, OpenAI says it\u2019s developed a more capable model that\u2019s able to deliver on the promise of AI agents.<\/p>\n

This story was updated with more information.<\/em><\/p>\n<\/div>\n