Openai, launch, agent of the general purpose of Chatgpt

Openai launches AI agent for new public purposes in ChatGPT, which the company says can complement a wide range of computer -based tasks on behalf of users. Openai says that the agent can automatically move in the user evaluation, create editing and slides -display offers, and code.

The tool, called Chatgpt Agent, combines many capabilities of the previous agent tools in Openai, including The operator’s ability to click on web sitesBesides Deep search capacity to synthesize information From dozens of websites to a brief research report. Openai says that users will be able to interact with the agent simply by paying Chatgpt in the natural language.

On Thursday, Openai puts a Chatgpt agent to its professional plans, as well as, and team plans. To activate the tool, users can determine the “agent mode” in the list of drops from ChatGPT.

Agent Chatgpt’s launch is the most daring attempts by Openai to date the converting Chatgpt to a product product that can take action and user disposal tasks, instead of answering questions. In recent years, Silicon Valley companies including Openai, Google and Perplexity have revealed dozens of artificial intelligence agents who promised to do so. However, this early version of artificial intelligence agents has proven that it is struggling through complex tasks, and it appears to be less convincing as products from VISION TECH executives around the agents of artificial intelligence.

However, Openai says that Chatgpt agent is more capable of its previous offers.

The new agent of Openai can access Chatgpt connectors, allowing users to connect applications such as Gmail and GitHub so that the agent can find information related to your demands. Moreover, Openai says that Agent Chatgpt has access to a station, and it can use application programming facades to access certain applications.

The form behind the Agent Chatgpt performs a performance on the latest model on many standards, according to Openai.

The company says that the Chatgpt customer model records 41.6 % in the last humanity exam (Passing@1), a difficult test consisting of thousands of questions across more than a hundred topics. This is the weakness of Openai’s O3 and O4-MINI in the test.

On Frontiermath, one of the most well -known standards of well -known mathematics, Openai says that Chatgpt agent records 27.4 % when he has access to tools, such as a plant to implement code. The previous modern result comes from O4-MINI, which recorded only 6.3 %.

Openai notes that he developed a Chatgpt agent with safety in mind, due to the fact that the product offers some newly discovered capabilities that may make it more dangerous in the hands of a bad actor. How ChatGPT agent is really capable, however, it is still seen.