4o Vs 4o-mini Vision API speed testing

[01jonny01]

Hi there has anyone test the response speed of 4o Vs 4o-mini when for vision particularly? I have done some testing and cannot see any response time improvements.

[Innovatix]

Based on my tests in the playground for image tasks, GPT-4o provides faster responses compared to GPT-4o mini. However, GPT-4o more confabulated also seen when counting the dog and for kitten GPT-4o mini do the same.

[supershaneski]

If I follow mini’s output, my package will be sent to the wrong address…

[01jonny01]

@Innovatix

the playground for image tasks, GPT-4o provides faster responses compared to GPT-4o mini. However, GPT-4o more conf

Thank you for this, I had similar results using the 4o-mini API, I also found that the 40-mini is more likely to refuse to analyse the image.

[grokthebuilder]

tried it for a very specific, complex and unique vision task for creating certain synthetic datasets, for my usecase it performs way worse than gpt4o in its output quality, kinda to be expected though, havent noticed any speed improvement

[01jonny01]

@grokthebuilder

specific, complex and unique vision task for creating certain synthetic datasets, for my usecase it performs way worse than gpt4o in its output quality, kinda to be expected though, havent noticed any speed improvement

Thank you for confirming this, I had a similar experience.