This is the launch of the most anticipated artificial intelligence of Openai in an increasingly fierce race

For months, the feeling was clear: OpenAi needed a blow to the table. GPT-4 had raised the bar, and GPT -4 -or reinforced that base with concrete improvements in performance and multimodality. But GPT –4.5 He did not convince: for many, it was a false step, With few real improvements and too many promises without compliance. Today, With GPT -5, the company seeks to recover impulse. The new model is now available and arrives in three versions.

Openai has profiled GPT -5 as its most complete model to date. His training puts the focus on three great pillars: understanding more precision instructions, managing longer contexts and working hand in hand with external tools. The ambition is not less: transforming the model into a kind of digital agent that not only responds, but acts, plans and adapts.

The clearest difference is how we interact with him. GPT -4 forced to choose: speed, vision or reasoning. GPT -5 breaks with that barrier. Now it is the model who decides in real time if he must respond instantly or stop to think. The result is a more fluid experience, without the need to configure modes or activate specific functions.

GPT -5 is already the new default model in the free chatgpt plan. It is the first time that anyone can access an advanced reasoning system without going through a box.

In payment plans, the model is deployed with more powerful versions:

• Chatgpt plus ($ 20 per month): access with extended limits
• Chatgpt Pro (200 month): unlimited access to GPT -5 Pro, a variant reinforced with more power, better answers and extended functions (voice, search, canvas, deep reasoning)

Enterprise, Team and Edu plans will receive it next week as a base model.

In development.

Images | OpenAI

In Xataka | I have tried the new OpenAI models. It has been a small odyssey with prize: I have a chatgpt at home