Openai unveiled the powerful Gpt-5 artificial intelligence model • DigiKala Meg

The expectations were finally over, and Openai unveiled its latest achievement, the advanced Gpt-5 artificial intelligence model. This advanced model, which has been developed today as a default and free model, is available to all ChatGPT users to provide a new experience of interacting with artificial intelligence.

Sam Altman, CEO of Openai, described the Gpt-5 as a big leap compared to previous generations. To describe this progress, he likened the experience of working with the Gpt-5 to the first iPhone equipped with a Retina display; A product that was almost impossible to return to older models after using it. Altman also pointed to the evolution of the company’s artificial intelligence, saying, “If GPT-3 was like a high school student and a GPT-4 student like a clever student, the GPT-5 conversation will first evoke a sense of talking to a doctoral degree.”

One of the key changes to the GPT chat interface is to delete the model’s choice. The Gpt-5 is designed with a smart system that automatically identifies complex questions and uses a more powerful version of reasoning to answer them. It is enough to use the term “Think Hard” in his or her command to activate this feature.

Although access to the GPT-5 is free to all users, free design users will have an unknown limit on the number of questions. Upon reaching this ceiling, the system will automatically place a weaker version called “Mini”. In contrast, Plus subscribers can use the full GPT-5 capabilities for longer. Developers will also access three versions of the Gpt-5, Gpt-5 Nano and Gpt-5 Mini at different prices via API. There is also a exclusive version of the Gpt-5 Pro for subscribers of the $ 200 month plan.

Alongside these technical developments, Openai has designed four new personality styles, Cynic, Robot, Listener and Nerd to personalize the answers. Users can also customize their chat environment with the desired colors.

The coding capabilities in the Gpt-5 have improved dramatically. Altman predicts this model will start a new era, “Software in demand.” “The Gpt-5 is the smartest model we have used for coding,” says Michael Troll, CEO CEO. This model not only identifies complex errors, but it can also complete multi -stage programs; The work in which previous models were incapable of. “

In the Swe-Bench Verified benchmark, which measured the ability to solve coding problems, the GPT-5 points earned 74.9 % and passed powerful models such as Claude Opus 4.1 and Gemini 2.5 Pro. In the GPQA Diamond doctoral exam, it also showed its competitors with 89.4 % score.

Safety and inaccurate data reduction has been one of the main priorities of Openai in the development of GPT-5. The company has attempted to minimize the rate of incorrect information by performing more than 5,000 tests. One of the interesting features in this regard is the model’s ability to detect its disability. As such, ChatGPT is honest to the user in cases where he is unable to do something. Another feature called Safe Completions is also designed to manage vague questions to prevent dangerous information. Finally, the accuracy of the model in the health field has also improved, and at the error rate of only 1.6 % in providing health information, it has become a more reliable tool.

Source: Openai