GPT-5 here is free for everyone

Two days after his first announcement Openai launches the long-awaited GPT-5. What’s more, you can Start using it todayEven if you are a free user. With GPT-5, the company promotes comprehensive improvements, claiming that the model is the best yet when it comes to coding, writing, safety, accuracy and more.

“GPT-5 is the first time that you really feel you are talking to an expert on any topic,” Sam, CEO of Openai (and Hypeman). “It reminds me when IPhone moved from that old giant pixel (screens) to the eye screen, then I went back to using one of those large pixels and I was like,” Wow, I can’t believe how bad we are. “

Altman said that GPT-5 will provide a unified experience for users, and the new model offers this promise. For the first time, Openai’s default view is a model of thinking, which means that the system is programmed to address complex problems by dividing them into smaller parts. Previously, if you want to force Chatgpt to use one of the thinking models in Openai, you should select the “Think Long” option from the claim tape. This means that most free users did not even know that Openai has more capable models. With GPT-5, the company has greatly simplified the ChatGPT experience.

On the side of the consumer of things, there are only three versions of the new model. One of these-GPT-5 Mini-burns only when free users run and decorate to the extent of the usual GPT-5. Another variable, GPT-5 Pro, is, as the name suggests, only available to the company’s subscribers . Regarding the theme of the query limits, users can use GPT-5 “more” more than those who have a free account, while professional customers can chat with GPT-5 as they want.

When it comes to thinking, GPT-5 is much faster than The former Openai of artificial intelligence. “It is so fast that I had a psychological experience of wondering, for example, is it really thinking enough? Then it gives a great answer,” Altman said. Perhaps most importantly, he suffers from fewer hallucinations, with Openai that the model provides more accurate answers than any of the previous thinking systems. We’ll see how GPT-5 responds to using the real world, but if Openai has made significant improvements here, it will be great; Halosa was usually a great weakness in thinking models, especially for traditional large language counterparts.

At the same time, Openai says that GPT-5 is the safest of artificial intelligence so far. For anyone, it includes a new feature called safe completion. “In the past, we are close to this type of the duo, if we believe that the claim was safe, we will be obeyed. If we think he was insecure, the model will refuse,” said Alex Biotel, Openai Safety Research Commander. “This has succeeded well, but as a challenge there can be a kind of claim that may be carefully confusing that can be confusing. So if someone says about the amount of energy needed to ignite some of the specific materials that can be an opponent trying to overcome safety protection and cause harm, or may be a student who asks the issue of science to understand the physics of this material.”

With safe completion, GPT-5 will try to give the most useful answer within the safety restrictions imposed by Openai. In difficult situations like one mud shown above, the model will only provide high -level information that cannot be used to harm anyone. “On average, the system is safer and more beneficial for users, and we think it will be much better,” he added.

In addition, when it comes to health questions, GPT-5 is better in concerns related to questions and suggests that the user ask him from the healthcare provider. It will also answer these claims more accurately, thanks to the ability to adapt to the level of knowledge of a person and geography.

In addition to everything else, Openai says that GPT-5 is the best coding model so far. He is also supposed to be a better writer, as the company is that Chatbot is better in translating your drafts into a “convincing, resonance” version.

Besides GPT-5, Openai adds a handful of new features to Chatgpt. To start, users can now choose a color for their conversations, with a few exclusive options available to pay customers. Openaii has also facilitated the connection of Chatgpt to Gmail, Google Calendar and Google. Once the connections are enabled, Chatbot will be known when to return automatically from your Google accounts; You will not need to determine anything before starting to chat. Openai will start offering this feature to professional subscribers starting next week, with other users available.

In the custom guideline, where you can write the system’s claim to modify how to interact ChatgPT with you, Openai offers a handful of pre -defined characters. Four options – satirical, robot, listener and natural memory – as part of the research preview, can be changed or disabled at any time.

Finally, not another, Openai issues an updated version of its advanced voice in the company . Openai says the tool is better in understanding the instructions and adapting the speaking style at the present time. As part of this change, Openai retires the standard audio mode. In practice, this means that the company can now provide a better audio experience for everyone because it does not need to return to the standard sound mode, which is not similar to the advanced voice, and thus in understanding the nuances of human speech.

If you are wondering about the whereabouts of this Openai on the road towards artificial general intelligence, Altman used to say this when asked about the topic. “I hate a kind of AGI, because everyone at this stage uses it to mean something a little different, but (GPT-5) is an important step forward towards the able models. We still miss something very important,” noting that GPT-5 cannot constantly learn on its own. “But the level of intelligence here, the level of ability, looks like a great improvement. Certainly, if I can return five years before GPT-3 and told me we have this now, I will be like this a large part of the road to something that looks like AGI.”