The Practice Room / The Desktop Apps / Episode 9
Which ChatGPT model should you pick?
Which ChatGPT model should you use?
The short answer
If you do not know which model you need, you do not need to choose. Default is the answer, and it is already ticked.
Where the picker is
In the desktop app, on the Work side, the model name sits next to the message box. Press the little arrow beside it and the list opens.
Default is first and already selected. It picks for you, per request, and that is a reasonable thing to let it do.
The only rule you need
Reach for the quick one when
- You would be annoyed to wait
- It is a fast factual question
- You are going to skim the answer
Reach for the thinking one when
- It genuinely matters
- You were going to read it properly
- A sloppy answer costs you something
You are not really choosing a model. You are choosing how much care the question deserves.
Reading the names
The rest are names like GPT-6 Astra. The number is the version. The word is the model.
Once you know that, the list stops looking like a wall of jargon and starts looking like what it is: a few generations of the same thing, with a name attached to each variant.
Do it yourself
- Open the ChatGPT desktop app and go to the Work side.
- Find the model name beside the message box, and press the small arrow next to it.
- Read the list. Notice Default is first and already selected.
- Leave it there. Ask something you actually need today.
- Only if the answer is sloppy on something that mattered, come back, pick the thinking one, and ask again.
That last step is the whole skill. You move for a reason, not in advance.
Check yourself
Where is the model picker?
On the Work side, next to the message box. Press the little arrow.
In GPT-6 Astra, what is the number and what is the word?
The number is the version. The word is the model.
You do not know which one you need. What do you pick?
Default. It is first, already ticked, and it chooses for you.
When it goes wrong
Your list does not match this page. It will not. The names change every few months. Read the number as the version and the word as the model, and a renamed list stops being a new list.
The answer is slow and you are waiting. You are on a thinking model for a question that did not need one. Go back to Default, or to the quick one.
There is no picker at all. You are on the Chat side. The model name sits beside the message box on the Work side.
Words used here
Go deeper
OpenAI publishes a page listing every model currently available and what each one is for, and it is the only place the names are reliably current.
What she says in the film
Which ChatGPT model should you use?
In the desktop app, on the Work side, the name sits next to the message box.
Press the little arrow.
Default is first, already ticked. It picks for you.
The rest are names like GPT-6 Astra. The number is the version, the word is the model.
If you do not know which you need, you do not need to choose.
Default is the answer. No permission required.
Book a session and we do it together on a call.
Book an hourAsk what did not make sense
Comments go through our own form and are read before they appear. Links are not allowed; describe it in words.
More from The Desktop Apps.

Claude Desktop: Chat vs Code, what actually changes
This is the Claude desktop app.

Where the settings are in Claude Desktop
You opened Claude, and you cannot find the settings.

Does Claude run on your computer?
Give the chat a job, and it does not run on your computer at all. It runs on a computer somewhere else, one th…
Get the next walkthrough.
One email when something goes up. The argument, the film and the term, nothing else. One click to leave.
Have a question now? Ask Mstimaj, she knows the archive.