GPT-5 is here. What's next?

AI is no longer science fiction — in Kazakhstan, it already helps businesses write marketing texts, doctors analyze images, and the government automate services. And now OpenAI is rolling out GPT-5 — the most talked about innovation in the world of technology. But the question is worth asking: is this a revolution or just another update that we will quickly get used to?

The long-awaited release, which has been rumored for months, has finally happened. GPT-5 has made a noticeable step in the convenience and quality of interaction with ChatGPT. The main “feature” is that the line between regular models and those that can reason has disappeared. Now the system itself decides when to use the fast version, and when to turn on a slower, but smarter mode. For paid users, this is already available, and free users will have to wait a little.

If you compare GPT-5 not with GPT-4, but with the first “intelligent” model o1 (released last year), the difference is obvious. o1 was a breakthrough — it could “think” before answering, solving much more complex problems. GPT-5 is a product brought to perfection. As Sam Altman aptly noted, it’s like the transition to a Retina display in smartphones: the picture became clearer, it’s more pleasant to use, but the world didn’t turn upside down.

In the GPT-5 demonstration, for example, they asked to create a web application for learning French. The model did a great job. But GPT-4o did almost the same thing — it just looked a little less “glamorous”.

There are, however, real improvements:

• GPT-5 reasons faster than the old “o-models”;

• it chooses when to turn on complex logic;

• the launch for free users hints at lower production costs and environmental friendliness;

• and, perhaps most importantly, it “hallucinates” and makes up facts less.

  
  

GPT-5 has achieved record results in a number of tests. But experts remind us: many benchmarks are already “saturated” — as if a high school student were tested with problems from middle school.

The bottom line is that GPT-5 really feels like a more “lively” and convenient tool. Ordinary users, far from technical nuances, will certainly feel it. But a feeling is not a revolution yet. After the real breakthrough that reasoning models gave, we are still waiting for the next step.

The question is when it will happen.