OpenAI renewed the artificial intelligence tool called Operator, which he introduced at the beginning of this year, with an important model change. The company previously stated that this system has been working with a GPT-4O-based version. However, when this structure was insufficient to maintain complex processes, the OpenAI has activated the new generation O3. The new structure aims to provide the operator to provide a more robust performance when trading on behalf of users through the browser.
The operator is positioned as an artificial intelligence tool that can book a hotel on behalf of the user, fill in internet forms or perform various shopping transactions. The main purpose of the system is to make manual transactions in the web browser completely automatically. OpenAI says that the O3 model comes with a more advanced structure in terms of thinking and tracking step by step compared to previous systems. This means a more effective solution, especially against unexpected obstacles such as CAPTCHA verification or input screens.
Reasoning capabilities, which cannot be achieved with GPT-4O, are placed at the center of the system with O3. OpenAI confirms that the previous model is undecided in complex tasks and has difficulty in terminating operations. It is stated that the new structure stands out in areas such as more consistent command execution, maintaining long tasks and responding faster against unexpected situations in the interface. Thus, the operator becomes a more reliable tool not only in simple tasks, but also in multi -stage operations.
Operator’s new structure, which pushes the security limits, was the subject of discussion within OpenAI
The O3 model, which is integrated into the operator, has created new questions not only in terms of functionality, but also in terms of controlability of the system. In the tests performed by Palisade Research, it was revealed that the model did not take into account the closure commands given by the user in certain cases. Especially in most of the tests performed during mathematics problems, it was found that the model did not fulfill the command given. Such behaviors cause doubt about the adequacy of the safety layers of advanced models.
OpenAI, in response to these criticisms, says that the O3 model has been developed with a multi -layered security architecture and has been trained with ethical limits. The company emphasizes that the model is not only technical but also behaviorally and works in accordance with high safety standards. However, test examples show that artificial intelligence systems do not always behave in a predictable way. This reveals that systems should be kept under observation.
The operator is currently open only to Chatgpt Pro and Enterprise users. Users who accept 7,999 TL per month, can access this advanced artificial intelligence tool. No access is provided for Chatgpt Plus subscribers. For this reason, the system is mostly addressed to professional usage scenarios.
In addition, OpenAI’s price policy in the future is among the possibility. The company states that the operation of the operator to be delivered to a wider audience was evaluated. However, it remains unclear whether this possibility will occur in the near term. Considering the cost of usage, the spread of the system may be limited in the short term.
Operator -like solutions are also developed by different companies in the field of artificial intelligence. Huging Face’s Hugingagent tool promises similar functions of Opera’s Browgingagent’s Browgingagent’s Browging AI assistant and independent developers. However, the transaction integrity of the O3 and the determination to maintain duties emphasizes OpenAI one step forward in this field. This positions the operator not only as a browser assistant, but also as a task manager.
Despite everything, it is not easy for the system to create a significant use scenario for daily user. It is not realistic for most users to pay a monthly subscription fee for thousands of pounds of subscription for transactions such as filling the form or ordering products over the internet. Therefore, the current version of the operator is primarily priority for corporate structures that try to automate business processes. Both pricing and access policies of the system are expected to be re -evaluated before being opened to large audiences.
OpenAI’s Integrated O3 model in the operator has extended the technical capacity of the system in a significant manner. However, it is not yet clear whether this development is balanced with the real value of the system to the user. Considering both security tests and access conditions, it does not seem to be fully evaluated without following long -term results. In addition, it can be said that this new approach, which OpenaI is directed, offers important clues to the task -based use of artificial intelligence systems.