OpenAI has officially introduced GPT-5, the new generation of its renowned chatbot. The model is expected to deliver state-of-the-art performance and will become the standard model for every user.
OpenAI has released ChatGPT-5, the long-awaited successor to GPT-4. The new language model aims to provide more accurate answers and better handle complex questions. The model replaces all previous versions and immediately becomes the standard for all users, including the free version.
In previous generations, users received different models, each with their own strengths. GPT-5 combines these into one system. Contrary to what GitHub leaked yesterday, we now know there are three models: ChatGPT-5, ChatGPT-5 Thinking, and ChatGPT-5 Pro. It decides itself when it can answer quickly and when extra reasoning steps are needed. This should improve the user experience.
Doctorate Level
“ChatGPT-5 is a significant leap forward in intelligence compared to our previous models,” reads the blog post. “It is a unified system that knows when to respond quickly and when extra reasoning steps are needed.” The model can perform a wide range of tasks, from building software applications to managing schedules.
Although ChatGPT has 700 million weekly users, it has been a while since a model was top of the class in terms of performance. That is now the case again. According to The Verge, Altman compares his models to students: “GPT-3 felt like talking to a high school student, GPT-4 seemed like someone from college, and with GPT-5 you finally feel like you’re speaking with a doctoral student.”
Faster and more Versatile
According to OpenAI, GPT-5 responds up to 50 percent faster to simple questions than GPT-4. For more complex tasks, such as creating an app or analyzing large datasets, the model automatically switches to a more extensive mode. You can choose from different personalities, including Cynic, Robot, Listener, and Nerd, allowing the tone of responses to be adjusted.
For developers, programming capabilities have been expanded. GPT-5 can set up complete web and app projects from a text description, including additional features and visual finishing. Integrations with Gmail and Google Calendar have also been added, allowing the model to manage appointments and prioritize emails.
Improvements and Limitations
The model hallucinates less often than previous generations: GPT-5 provided incorrect information in 4.8 percent of cases, compared to more than 22 and 20.6 percent with GPT-4o and o3. GPT-5 is also less inclined towards misleading or incorrect behavior.
OpenAI writes: “GPT‑5 is generally smarter and provides more useful responses in the fields of mathematics, science, finance, law, and more. It’s like having a team of experts for all your questions.”
Availability
GPT-5 is available to all ChatGPT users. Free users have limited access and switch to GPT-5 mini after their limit. Plus and Pro subscribers receive larger usage limits. Pro users also have access to GPT-5 Pro, which performs longer reasoning steps. Team, Enterprise, and educational users will gain access to the new model this week.