I get this question frequently and the only way for me to respond to this is by asking “Are you measuring it the wrong way?”

I keep seeing reports saying “GPT-5 isn’t exceptional” or “progress is slowing down compared to GPT-3 and GPT-4.” And honestly, I doubt that’s the right way to look at it.

Lets replay this using simple logic:

When people first used GPT-3 and then GPT-4, the difference felt huge. Why? Because most users were asking simple things like basic conversations, translations, or easy reasoning tasks. GPT-4 handled these with a lot more fluency, so it felt like a massive jump.

But now, when the same people test GPT-5, they’re still asking the same simple questions. And here’s the truth: if GPT-4 was already giving you a 99% accurate answer to “2 + 2 = 4”, GPT-5 giving you 99.5% accuracy will not feel earth-shattering.

It’s like comparing a PhD scholar and a 3rd grader solving “2 + 2.” Both will say “4.” The difference will not feel dramatic because the problem itself is too simple.

But when you put that PhD scholar on a complex math problem, that’s where you will really see the difference.


Similarly, GPT-5 is already showing signs of solving far tougher problems that earlier models struggled with. For example (I will leave the links in the comments sections for you to read these examples in detail):
1️⃣ Medical reasoning: Outperformed doctors in complex multimodal diagnosis tasks.
2️⃣ Advanced coding: Built and debugged complex apps, with design sense in UI.
3️⃣ Math & logic: Scored near-perfect accuracy on advanced math and coding benchmarks.
4️⃣ Everyday challenges: In head-to-head tests, GPT-5 gave more realistic, useful, and empathetic answers than GPT-4.

So maybe the issue isn’t that AI progress has slowed. Maybe it’s that we’re still testing it with the wrong benchmarks. Instead of asking GPT-5 the same old questions, we need to watch how it performs in real critical use cases like science, research, healthcare, and business operations where complexity truly tests its limits.

That’s when the leap will be obvious.