OpenAI launches a general purpose agent in ChatGPT

Openai says the company says the company says a new general purpose AI agent that can perform various computer-based tasks on behalf of the company. Openai says the agent can automatically change the calendar of a user, create editing presentations and slide shows and can activate the code.

Chatgpt agent means a number of opportunities from the vehicle, including Openai’s previous agent The ability to click on the operator’s websitesAs well as Facilities for information synthesis of deep research to a short research report from dozens of websites. Openai says that with the agent of users, they will be able to interact with the request of ChatGPT in a natural language.

Thursday, Openai, Subscribers spread the wildlife for subscribers to Pro, Plus and Team Plans. To activate the instrument, users can select “Agent mode” on the Dropdown menu of ChatGPT tools.

The start of the Chatgpt Agent represents the most courageous attempt to turn to an agent product that can use ChatGpt’s questions to users, to use the action and download tasks for users. In recent years, silicone valley companies, including Openai, Google and confusion, presented dozens of AI agents that promise to do it. However, this early version of AI agents proved to fight complex tasks, and the EU agents seem to be less coercive than the last vision technology employees.

He said: “Openai says that Chatgpt is more capable of its previous victims.

Openai’s new agent can include ChatGPT connectors, which allows users to close applications such as Gmail and Github, in addition, Openai says that the Chatgpt agent can use API to a terminal and access to certain applications.

According to Chatgpt Agent, Chatgpt Agent, Openai offers the most modern performance in several criteria.

The company, 41.6% of the Chatgpt Agent Model, 41.6% (Pass @ 1) is a difficult test consisting of thousands of questions between a hundred and more than a hundred subjects. In the test, O3 and O4-Mini’s O3 and O4-mini increases about twice.

In Frontierath, Openai, Chatgpt Agent, which is one of the most difficult-known math benchmarks, Chatgpt agent considers 27.4% when the code is included in the Terminal for the execution. The previous the most modern score comes from O4-Mini, which hit only 6.3%.

The Openai notes that this is a security with security, mainly, according to some new opportunities that can be more dangerous in the hands of a bad actor, Chatgpt agent. The strong chatrpt agent remains to look really.