>>>ANSA/OpenAI launches Gpt-5, a new step toward general AI

(by Antonino Caffo) Expected for months and promised by mid-August, OpenAI has not He betrayed the timing and revealed GPT-5. This is the new artificial intelligence model of the American startup, which powers the popular ChatGpt chatbot. A major update not only from a technical point of view, since it should be the last before the agi, the general artificial intelligence, in able to reason like a human even in complex activities. It is available immediately, even for free users. As per OpenAI's official note, Gpt-5 is a new reference in terms of accuracy of responses and less incidence of "hallucinations", the errors that can lead to I meet artificial intelligence. But not only that, "GPT-5 is the OpenAI's best model ever built for questions related to health," the note reads. "It can help to understand the results, to ask appropriate questions submit to specialists and evaluate options at the time to make decisions. Gpt-5 also offers more precise answers and reliable and adapts to the context, level of knowledge and to the user's geographic location, thus providing more accurate answers safe and useful in a wide range of scenarios." The model is unique but divided into three commercial 'offers'. The standard version, Gpt-5, is optimized for reasoning and for tasks that require multiple logical steps. Gpt-5-mini is a lighter and lower cost declination, designed for consume less. Gpt-5-nano is instead designed for speed and, according to OpenAI, it is the choice for applications that require low latency, i.e. a minimum time between a request and the AI response. OpenAI also focuses on personalization. On the one hand four personalities to choose from, Cynic, Robot, Listener and Nerds, who answer in a concise and professional manner, thoughtful, encouraging, slightly sarcastic. The options are optional and adjustable at any time in the ChatGpt settings. Then the integrations with Gmail, Google Calendar and Google Contacts. Once the connections are enabled, ChatGpt will be able to read data in apps and produce responses contextual based on information. OpenAI developed its first model, Gpt-1, in June 2017. 2018. At the time, it was considered revolutionary, with its 110 millions of parameters, and the ability to generate text coherent and sensible. Just a year later, in February 2019, Gpt-2 arrived, so powerful in text generation that OpenAI decided initially not to release it for fear of misuse. The real turning point was Gpt-3, released in June 2020. It could translate languages, generate code and perform a wide range of activities. This was the basis of ChatGpt, published almost three years ago. In March 2023, Gpt-4 was introduced, which showed noticeable improvements in depth of understanding, in reasoning and producing more accurate answers and faded. Before GPT-5 it was GPT-40's turn, in May 2024. The "o" stands for "omni," to emphasize its nature. unified multimodal. Unlike previous models, which handled text, images and audio separately, Gpt-4o integrates all these modalities in a single neural network.