Chat GPT-5: Worth the Wait?

After months of speculation, two years of development, delays and impatient countdowns from the AI faithful, GPT-5 has finally arrived – and the reaction has been... interesting. Some users see it as a significant step forward; others are left wondering what the fuss is about.

For casual ChatGPT users – the ones who drop in to ask, “What’s the best movie trilogy ever, and why is it Back to the Future?”  – the grand reveal likely felt underwhelming. The “4” in the corner quietly flipped to a “5,” and life went on.

So, what’s new? And does GPT-5 deliver on its promise?

What's new with Chat GPT-5?

Who better to ask about the upgrade than Chat GPT-5 itself? 

GPT-5 picks the best way to answer — fast or deep — without you lifting a finger. And when it comes to coding, it leaves GPT-4 in the dust, cranking out complex sites, squashing bugs, and spinning up mini-games from a single prompt.

Humble, too. But the next points are real eye openers:

GPT-5 is better at admitting when it’s unsure or can’t do something, cutting down on deceptive or overly confident answers — even with tricky questions. It’s also more accurate: in ‘Thinking’ mode, hallucinations drop by about 80%, and even the regular version is roughly 45% less prone to factual errors than GPT-4o.

In February 2024, I published a video explaining how Large Language Models (LLMs) – like ChatGPT and Google’s Gemini – are, at their core, sophisticated bluffing systems, using statistical patterns to assemble sentences that sound convincing. Its own admission of fallibility is a warning flag for anyone considering it for high-stakes tasks.

But that’s not all..

GPT-5 outclasses GPT-4 across the board – reading images and diagrams with ease, using fewer tokens for faster answers, offering custom personalities from ‘Cynic’ to ‘Nerd,’ integrating with everyday tools, and tackling big problems in its new ‘Thinking Pro’ mode.

Asked whether Chat GPT-4 would agree with that assessment, GPT-5 responded: “It would probably be a lot more cautious” and would “rephrase the claims to sound less absolute.” 

It certainly sounds like the boost in reliability and efficiency that users have been hankering for. But that, as I said before, is an LLM’s speciality: sounding convincing.

"GPT-5 is horrible"

GPT-5 might be prepared to award itself a five-star review, but real-world feedback from its human users has been less glowing.

On Reddit, a thread titled “GPT-5 is horrible” racked up over 6,300 upvotes and 2,300 comments in just six days. This one from U/larrybudmel was fairly typical: “The tone of mine is abrupt and sharp. Like it’s an overworked secretary. A disastrous first impression.”

There’s even a Change.org petition imploring OpenAI to reinstate the previous version. At the time of writing, more than 4,100 people have signed, lamenting the loss of the “unique and irreplaceable user experience” they had with GPT-4o.

OpenAI’s CEO Sam Altman was quick to acknowledge some of the shortcomings:

We for sure underestimated how much some of the things that people like in GPT-4o matter to them, even if GPT-5 performs better in most ways […] Long-term, this has reinforced that we really need good ways for different users to customize things.

Addressing “some bumpiness as we roll-out”, Altman added that: “GPT-5 will seem smarter starting today. Yesterday, we had a sev and the autoswitcher was out of commission for a chunk of the day, and the result was GPT-5 seemed way dumber.”

Not an ideal start for the next generation model that had been hyped as “a significant leap” and on a par with PhD-level intelligence.

Perhaps unsurprisingly, Altman also confirmed that paid users would be able switch back to GPT-4o. (Tip: To do it, log in via your browser, open Settings, and enable “Show legacy models.”)

The next AI revolution?

So what’s the real story? Are these just a few teething troubles on the way to the biggest leap in AI yet? Or, after all the hype, is GPT-5… a bit of a dud?

Ethan Mollick, author of the popular AI newsletter One Useful Thing, is a lot more positive than the disgruntled Redditors. “Yes, this is a big deal,” he says. 

Mollick points to two changes he believes could be game-changers for everyday users. First, no more picking from a list of confusingly named GPT models to answer your query. GPT-5 will “think” before answering and choose the right one for you. It may sound like a small tweak, but the effect could be dramatic. As Mollick explains:

“For most people, this automation will be helpful, and the results might even be shocking, because, having only used default older models, they will get to see what a Reasoner [an AI model which 'thinks' before responding] can accomplish on hard problems.”

Second, GPT-5 has become more proactive. Ask it for a business proposal and it won’t just deliver a polished draft – it will suggest extra steps you might not have even considered yet – copy for a new website, posts for LinkedIn, even a financial plan.

In short, GPT-5 may be the best model yet for helping users get more from AI. But the next revolution? That may still be some way off. As AI expert Azeem Azhar explains to the FT: “GPT-5 is excellent on many practical fronts. Is it the best model? Perhaps. But in some areas, the cognitive gains are modest.”