OpenAI unveiled its eagerly anticipated ChatGPT-5 on 7 August, describing it as the firm’s “smartest, fastest, most useful, model yet”.
The upgrade is available to all 700 million users for free, though there are usage limits for those not paying.
Feedback from users has so far been mixed, with some welcoming the advancements and others longing for a return to previous models, claiming GPT-5 has less “personality”.
Here, Cayman-based AI thought leader Tamsin Deasey-Weinstein, also senior director at UCCI, gives us her thoughts on the revamp:
ChatGPT-5 promised the same as every other ‘world-changing upgrade’. A larger context window (which means you can give it more data to pull information from), faster and more efficient responses, improved coding and reasoning abilities.
Then there was the big one. GPT-5 was said to be a huge leap toward the elusive AGI (artificial general intelligence).
AGI is the kind of AI that can think, learn and act across many different tasks like a human.
It can’t just find you your favourite product online. It can find it, analyse competitor pricing, choose the best, order it, pay and have it delivered to your door. All without you clicking a button. You see the difference?
How was it doing this? The headline change? It is now one unified model. Gone are the confusing options, GPT4, o3, o1, mini versions and many more.
ChatGPT-5 now chooses the best mode for you automatically, drawing on your request, past results and even usage limits (switching to a smaller version if needed).
We no longer even see the different models. In theory this also allowed ChatGPT-5 to stack, to make it more ‘agentic’.
It could stack its reasoning models with its image generation models and even its agent model to do end-to-end tasks as described above.
For casual users, this is a win. The majority of casual users never switched models anyway, so GPT-5 will give them a massively more intelligent answer.
I expect it’s been life changing for those. No longer do you have to learn about every model and know their nuances. AI will do that for you!
But for power users, it’s not so simple. Losing manual model selection means losing personal control over style, depth and performance.
Where you once might have chosen o3 for deep research or a creativity-tuned model for brainstorming, now the AI decides.
For me, the output feels like a downgrade. I have also tried a variety of ways to get GTP-5 to stack tasks without my intervention and its failed miserably. Every single time.
Why the change OpenAI? Likely cost and efficiency. I believe that OpenAI may have found that most people defaulted to the smartest models for even simple tasks, expensive for them, unnecessary for the task.
I am pretty sure that most people also stuck to a single model out of habit and a lack of knowledge, getting subpar results and leaving dissatisfied. So OpenAI have chosen to cater for mass market rather than power users.
The good news; ChatGPT Plus, Team and Pro users can still reenable legacy models in settings, and advanced modes like Agent or Deep Research remain, if you know where to look.
The even better news, models change daily. OpenAI will see the faults and will be working hard to fix them. This time next week the model will be improved. This time next month it may be a whole different model.
The bottom line is that ChatGPT is still the leading frontier model for a wide range of tasks. The new launch may not be perfect for everyone, but it’s a good step forward for many.
My advice to any and all, try it for yourself. You have nothing to lose.
Deasey-Weinstein is co-chair of the Cayman Islands AI Steering Committee (Education) and steering committee member of the Cayman Islands Artificial Intelligence Association.
She, together with her UCCI colleague, dean of humanities and social sciences Dr Genève Phillip-Durham researched and wrote ‘Harnessing artificial intelligence for educational reform and workforce development: The case of the Cayman Islands’.
The paper was published in the prestigious ‘The Round Table: The Commonwealth Journal of International Affairs’ earlier this year.
