Which is Faster GPT-4, GPT-4-Turbo, or GPT-3.5-Turbo? - The Answer May Surprise You!

[@ronalddhs3726]

Thanks Gary. Just one question: Python tutorial, when? :) Please.

[@jordancave6987]

Awesome stuff Gary. If you have the time, I would love to see you do some this testing for gpt vs bard.

[@themenon]

Would love to see an API tutorial from you on OpenAI API + Python

[@JoelJosephReji]

Yes, would like to +1 this. I would also love to have an extension of this idea where we get to see how local models can also be accessed via APIs in Python. (especially dealing with all the different parameters and flags, etc)

[@scrollop]

Yes please, if you could do a tutorial on how to access the API via python. Thanks for the interesting video.

[@HeadsetHistorian]

A performance benchmark of LLMs is the most Gary thing ever.

[@GaryExplains]

🤣

[@seniorp9444]

I’m seeing a lot of posts on X that GPT4 is getting lazy and doesn’t want to complete tasks. Several times in the last couple weeks Ive asked it questions about things like “give me the plot of that new show “ and it says it doesn’t know anything about that and can’t help. I have to say “search the internet and give me an answer “. Trying to save compute resources? It’s been pointed out that this type of behavior just leads to more queries for the AI as nobody is going to believe GPT4 really can’t do simple things it used to do.

[@xeon2k8]

well tbf, that's a really stupid question to ask to an AI, so maybe it was encouraging you to look it up yourself.

[@MeinDeutschkurs]

Nice. Isn’t gpt-3.5-turbo only a redirect to another (current) main model? Maybe this leads to some kind of sub queue. If you call the “real” model behind of it, it seems to be quicker.

[@RichardGetzPhotography]

Are there any benchmarks on how many tokens per second it can process?

[@StuPid-q1o]

Thank you. Python api tutorial please 🙏

[@ThePianist51]

Answer is simple: less tokens, faster performance. ^^

[@marcusk7855]

The Q-star version is probably the fastest but we'll never get our hands on it. LOL.

[@jeffgrubich8422]

You spelled Blackhawk wrong.

[@toungeenaamoo]

gpt4 on bing is so fqing slow

[@RohiGupta]

very interesting video, pls make python video too..