How to Choose the Right ChatGPT model

Learn about real differences between the model options on a real world research task

Sep 29, 2025

Comparing ChatGPT’s Model Options

With ChatGPT 5’s launch and recent updates, paid users can now choose between 4 models that trade off speed and depth: Instant, Thinking (Standard and Extended), and Deep Research.

I thought this was a good opportunity to compare how the available models perform differently.

How I tested it
I asked each model the same prompt about a real knee condition I have:

“Help me learn about patellofemoral pain syndrome and what exercises I should do to help.”

This is a real need I have been seeing a doctor and PT about so I wanted to see how ChatGPT would do.

My evaluation
The Standard Thinking model or Deep Research had the best results for the task. The Standard model produced a much more detailed plan than the Instant model. It was worth the 30 sec wait.

Meanwhile, the Extended Thinking model output took longer for a similar result to the Standard model and paled in comparison to Deep Research. The Deep research guide gave me accurate and helpful guidance and still completed in a respectable 5 minutes.

Caveats
Given the non-deterministic nature of LLMs, there is some variation even when using the same model with the same prompt.

I tested this with OpenAI models but the same types of reasoning, non-reasoning, and deep research models exist for Claude (Anthropic) and Gemini).

Takeaways
Companies are adding and updating models so fast it is useful to keep testing them. This was just one example so I’ll be trying it out on some more use cases in the future.

Would love to know anyone else’s experiences. Has anyone found a strong use case for Extended Thinking?

See for yourself

Here are the screenshots of the model outputs.

ChatGPT 5 Instant Model