OpenAI introduces new AI model GPT-5

On August 7, OpenAI unveiled the next generation of its neural network — GPT-5. The updated model is said to deliver more accurate reasoning, improved long-term memory, and a significant reduction in incorrect answers. The presentation was streamed on the OpenAI website.

“GPT-5 is smarter in every way, providing useful answers in mathematics, natural sciences, finance, law, and other fields. It’s like a team of experts ready to help you with any question,” an OpenAI representative said.

GPT-5 will be available to all ChatGPT users, though free users will face a limit on the number of requests. After reaching that limit, ChatGPT will switch to a simpler model. Through the API, users will be able to choose from three versions: GPT-5, GPT-5 Mini, and GPT-5 Nano. There will also be options to customize the chat color and select one of four communication styles: Cynic, Robot, Listener, or Nerd.

According to OpenAI, GPT-5 will fabricate fewer facts and more readily admit when it cannot answer. The company says 5,000 hours of security testing went into its development.

“The impression will be such that you will not want to go back to previous models. If GPT-3 is a schoolboy, GPT-4 is a college student, then GPT-5 is like an expert with a doctorate,” said OpenAI CEO Sam Altman.

The release of GPT-5 had been postponed several times. Initially planned for mid-2024, it was later moved to mid-2025, before being rescheduled for August. During this time, ChatGPT’s user base continued to grow. In April this year, Altman said weekly active users had reached 500 million, and OpenAI expects this figure to soon hit 700 million.

OpenAI, founded in 2015 by Elon Musk and other entrepreneurs in San Francisco, specializes in developing artificial intelligence technologies. Musk left the company in 2018. In November 2022, OpenAI launched its ChatGPT chatbot, capable of generating text, engaging in conversation, identifying programming errors, writing poetry, creating scripts, making predictions, and offering advice.

Nate Soares, a former engineer at Google and Microsoft and now head of the Machine Intelligence Research Institute, has previously estimated that the probability of artificial intelligence leading to humanity’s extinction is “at least 95 percent.”