Thanks for the comparison! Maybe a good idea to do a bit more in depth comparison, like changing the prompts a bit here and there and comparing outcomes (sort of a sensitivity analysis) and use some kind of (probably subjective) metric. Also it would be a good idea to just record video and play that through the API to really ensure the video and audio input is the same.
Wow, with 15 comments it feels so cozy here! Hi Greg, great comparison, you are so cool with those dev tools of yours 😂. Subscribing, hoping for more in depth comparison ;) actually will try to recreate and do it myself.
I've also compared both in api pro vision for my android app. I found open ai/gpt 4 response quick/fast and accurate than gemini. But yes, gemini is still "new". Your result is very clear.
nice demo , i think it would be greate to enlarge the dataset to see the difference. from your demo i can say that chatGPT 4.0 gives the result that any person would expect, but i think with more different dataset Gemini or with Gemini Ultra could be better than chatGPT 4.0. Great Work
Hi Greg! Once again great work. Creative setup to compare the two models that way. As I said in my other comment, I spent quite some time today with lots of GPT tests. Kind of complex ones too. Too much to put into that comment. Surely would like to get in touch with you. I'm impressed by what you do and the way you present it. A big 👍and Thx from me, Tom!
I guess we’ll have to wait for ultra to have a fair comparison, also these are just toys and sandboxes… Reality comes out when you have to apply grounding, policies and security to your environment, and that’s something you can’t build on OpenAI
"That's the demo think whatever you want to think" has to be the best outro I have heard. 😆
Results are so clear! Super outro 😂
Merci pour ça! Il faudrait refaire la même expérience encore, aujourd'hui, pour mesurer les changements dans les performances respectives.
Thanks for the comparison! Maybe a good idea to do a bit more in depth comparison, like changing the prompts a bit here and there and comparing outcomes (sort of a sensitivity analysis) and use some kind of (probably subjective) metric. Also it would be a good idea to just record video and play that through the API to really ensure the video and audio input is the same.
Wow, with 15 comments it feels so cozy here! Hi Greg, great comparison, you are so cool with those dev tools of yours 😂. Subscribing, hoping for more in depth comparison ;) actually will try to recreate and do it myself.
Very impressive work Greg!
Hahaha! I love these demos! It feels so natural and you let them just be. I am looking forward to see if Gemini Utra beats GPT4 in a demo like this!
but 99% people will never have access to the ultra model? while 99% people have access to GPT4
I knew that someone would definitely do what I wanted to see! Thanks Greg! )
I've also compared both in api pro vision for my android app. I found open ai/gpt 4 response quick/fast and accurate than gemini. But yes, gemini is still "new".
Your result is very clear.
nice demo , i think it would be greate to enlarge the dataset to see the difference. from your demo i can say that chatGPT 4.0 gives the result that any person would expect, but i think with more different dataset Gemini or with Gemini Ultra could be better than chatGPT 4.0. Great Work
Hi Greg!
Once again great work. Creative setup to compare the two models that way. As I said in my other comment, I spent quite some time today with lots of GPT tests. Kind of complex ones too. Too much to put into that comment. Surely would like to get in touch with you.
I'm impressed by what you do and the way you present it. A big 👍and Thx from me, Tom!
I really appreciate it!! Reach out - hi@greg.technology and let’s have a chat!
@@greg.technology great. Just did so.
not a fair comparison, you should compare ultra to gpt4, no?
Why are you saying "hey Gemini" and "hey gpt". Shouldn't you just be saying "what am I playing?"
Just a way to make it more obvious “who” (what) I’m talking to :)
I guess we’ll have to wait for ultra to have a fair comparison, also these are just toys and sandboxes… Reality comes out when you have to apply grounding, policies and security to your environment, and that’s something you can’t build on OpenAI
Gpt 4 is better
you should ideally compare with ultra. pro is equivalent to gpt3.5