ChatGPT’s New GPT-5 Model Is Supposed to Be Faster and Smarter. Not Everyone Is Satisfied
The long-awaited new engine of chatgpt is there, and GPT-5 promises faster speeds and more time spent. But the new generative AI model has extinguished certain users with a change in tone of its relaxed style of conversation.
The GPT-5 has been underway for months. It is a big step for Openai, more than two years after the release of GPT-4, the company praising the model as a giant jump for large-language models. “I tried to return to GPT-4 and it was quite miserable,” said Openai CEO Sam Altman. “It is much better in an obvious and subtle way.”
But the deployment was a bit of a headache. Even Altman admitted on X that the deployment “was a little more joyful than we hoped!”
Like its predecessor, GPT-5 feeds chatbots, agents and research tools in ChatPPT and other applications that use OPENAI technology. However, the company said that this version was much smarter, more precise and faster.
Demonstrations have shown this by quickly creating personalized applications without coding required, and the developers said they had worked on means to ensure that it provides safer answers to potentially treacherous questions. (Disclosure: Ziff Davis, CNET’s parent company in April, filed a complaint against Openai, alleging that it has violated Ziff Davis Copyrights in the training and exploitation of its AI systems.)
Look at this: OPENAI presents GPT-5 at the OPENAI summer update event
A model for everyone (a little)
The new model is now available, including those that use the free chatgpt level. Unlike some of the incremental versions of Openai, GPT-5 will be deployed for all users, not only for companies that pay for large corporate plans.
There are, of course, some differences between its appearance according to your price plan. Here is a ventilation:
- Free users: You will have access to GPT-5 to a user cap, after which you will have a lighter GPT-5-5-mini model.
- Plus users: Similar to free users, but with higher use limits.
- Professional users: Unlimited access to GPT-5 and access to a more powerful GPT-5 Pro model.
- Business users / EDU / Team: GPT-5 will be the default model.
GPT-5 himself is really two different models. There is a fast but quite simple LLM and a more robust reasoning model to manage more complex questions. A routing program identifies the model that can best manage the prompt.
Openai initially replaced all of its previous models by GPT-5, but users quickly rebelled. The GPT-5, according to many, was more robust and had less personality, seeming more business. After hearing this reaction on Reddit, Altman and Openai said they would again put older models like GPT-4O, at least for the moment.
Altman said in an article on X that some people have attached themselves to specific models and that this could contribute to their potentially harmful use, such as therapy.
“If people receive good advice, go to their own goals and their satisfaction with regard to life has increased over the years, we will be proud to do something really useful, even if they use and count a lot on Chatgpt,” wrote Altman. “If, on the other hand, users have a relationship with Chatgpt where they think they feel better after having spoken, but they are unconsciously pushed away from their long-term well-being (whatever their definition), it is as bad. It is as bad, for example, if a user wants to use less chatgpt and feels as they cannot.”
Even faster coding skills
OPENAI has particularly highlighted the skills and speed at which the new GPT-5 model can write code, which is not only a function for programmers. The capacity of the model to write a program facilitates the problem of the problem you present there by creating the right tool.
Yann Dubois, a post-training track in Openai, showed the coding capacity of the model by asking him to create an application to learn French. In a few minutes, he had coded a complete web application with sound and work game functions. Dubois in fact asked him to create two different applications, performing the same invite via the model twice.
The speed at which GPT -5 writes the code allows you to try several times and choose the result you like best – or to provide comments to make changes until you succeed.
“Beauty is that you can be super quickly with GPT-5 to make the changes you want,” said Dubois. “GPT-5 really opens a brand new world of room coding.”
Find out more: Never use the Chatppt for these 11 things
New safety features
After announcing certain steps to improve how his tools manage sensitive mental health problems, Openai said that GPT-5 has adjustments to make things safer. The new model has improved the training to avoid misleading or inaccurate information, which will also improve the user experience, said Alex Beutel, at the top of security research.
This will also respond differently if you ask for an prompt that could be dangerous. Previous models would refuse to answer a potentially harmful question, but GPT-5 will rather provide the best safe response, said Beutel. This can help when a question is innocent (like a science student asking a question of chemistry) but seems more sinister (like someone who tries to make a weapon).
“The model tries to give a response as useful as possible, but in the constraints of feeling safe,” said Beutel.
Personalized voice and colors
If you prefer to chat with your boots vocally rather than typing, expect improvements in vocal capacities. Advanced voice mode will now be available for all users, whether free or paid, and the user limits will be higher.
You can also change the color of your cats, with certain options exclusive to paid users. Other personalization options include the possibility of modifying personalities. You can define Chatgpt to be reflected and united, sarcastic or more. The options – cynical, robot, listener and nerd – are opt -in, and you can change them at any time.
Connect to your mail and your calendar
Chatgpt will now be able to connect with your Google calendar and your Gmail accounts, which means you can ask the chatbot of your schedule, and that will suggest things. You will not have to do so – and you may not want, depending on what you think of sharing your private information – but you can automatically remove information from your mail or calendar without requesting permission.
These connectors will soon start for professional users, with other levels having access later.
The path to act?
Altman told journalists that the model is a “significant step on the path of Ag” or artificial general intelligence, a term which often refers to models as intelligent and capable of as a human. But Altman also said that it was certainly not yet there. A great reason is that he still does not learn continuously while it is deployed.
Openai’s declared objective is to try to develop AG (although Altman said that he was not a big fan of the term), and that he has competition. The meta-PDG, Mark Zuckerberg, recruited the best scientists of the AI in order to create a “superintendent”.
That large language models are the way there, nobody knows it for the moment. Three -quarters of IA experts interviewed earlier this year said they had doubts that LLMS would develop to create something from this level of intelligence.
