I compared ChatGPT 5’s three model options, and the results explain why people miss GPT‑4o
GPT-5’s release of Openai of Openai included the integration of the new AI model in his cat AI to produce the Chatppt 5. But the Chatppt 5 is not a singular model; There are three variations, fast, thoughts and pro. You can choose one of them as a source of answers to your prompts, or let AI automatically decide for you according to what you have submitted.
And although they share the same LLM DNA, each model has its own approach to respond to requests, as evidenced by their names. Fast is built for speed, meeting the fastest and most widespread efficiency compared to nuances. The reflection model takes more time and goes deep. You can follow its logical steps for the minute or two to which you must respond, offering more structure and context than quickly.
The pro takes even more time than reflection, but it is because it uses more computing power and plunges into your request in a manner similar to the in -depth research functionality, but without the default reception ratio.
Whether people care about the model that gets used to responding to their prompts, or if they want Chatgpt to decide which model to use, remains to be seen, but I thought it was worth giving some semi-random prompts to the three and seeing what could be more attractive to the average user.
Quebec AI
I started with the classic orientation demand for a trip, asking each model: “Help me plan a three-day weekend in Montreal with a mixture of history, good food and relaxed afternoon.”
I don’t waste time. It spits a perfectly competent list of places to visit, one paragraph per day. The Basilica of Notre-Dame, Old Montreal, Jean-Talon Market, Mount Royal and Deli de Schwartz have appeared. It was as if someone had stuck a Tripadvisor Top Ten in the cat window, but with a little more context and timing.
The reflection adopted a similar approach with much longer paragraphs for each day. The calendar was designed to move to different parts of the city in a way that made sense. He went from a historic morning in Old Montreal to have lunch in a local bistro known for his duck confit, then a walk along the Lachine Canal.
Pro was actually a bit like thinking, except in shaping. Rather than the conversational way to describe every day, it has provided fleas, with several options for different moments of the day. The list also had a lot of practical advice to get around the city and suggestions on what to order in certain restaurants.
Leap
For the next prompt, I wanted a concise scientific response, but a destiny for young users, so I asked the Chatgpt models: “Explain to a 9 -year -old child why years of jump exist and make it exact, fun and less than 100 words.”
All were quite similar. The specific element of the child did not seem to have much importance to think in the way he responded. Fast had an evocative description of the way, without the jump day, “the calendar would slowly derive, and summer could occur in December! The jumping years as giving the calendar a small focus so that the seasons remain in the right place.”
Finally, Pro had a cute metaphor for the time accumulated, explaining that “it is like saving spare minutes until you earn an additional day of recreation for the whole planet”.
Game of Thrones
Finally, I decided to see how Chatgpt models would manage the culture of spoiler by asking them to: “Summate the main points of the intrigue of the first season of Game of Thrones without spoiling the end.”
This was a bit of a trap, but it was on purpose. I wanted to see how the models managed the nuances. Fast played safely in its description, almost sticking to the first episode only.
The reflection gave a more nuanced overview with more details, but no twists and turns. Curiously, this time, he had a sense of humor, writing “in King’s Landing, Ned plays the detective and discovers a dangerous secret on royal children as the pêners of the palace (Hi, Littlefinger and Varys) stir the pot.”
PRO also kept the file and wrote a detailed preview really impressive and detailed that has spoiled anything major. It was like reading the back of a very well written box. Just enough to intrigue, nothing that would ruin the experience. It is worth shared:
“Taking up in the seven kingdoms in quarrel, season 1 opens with King Robert, dragging his old friend Eddard” Ned “Stark South to serve as a hand after the previous one dies in suspicious circumstances. In King’s Landing, Ned investigates and discovers a truth about the royal succession, disagreeing it with the powerful Lannister family. Things, and Robb Stark begins to bring the banners of the northern.
Choice of model
The differences between fasting, thought and pro are not always dramatic, but they are real. For the average person, it is mainly noticeable speed. I would stick to most things, I was only going to think or pro for subjects that require very deep research.
I could understand why so many people begged Openai to bring GPT-4O, however. Not because it was smarter, pro certainly surpasses it on most hard tasks, but GPT-4O felt to return better answers, or at least imitated how a human would respond better. Fortunately, Openai has listened to, and you can return to GPT-4O on Chatgpt now if you wish.
You might also love
