OpenAI GPT-4o API Explained | Tests and Predictions
HTML-код
- Опубликовано: 19 июн 2024
- OpenAI GPT-4o API Explained | Tests and Predictions
👊 Become a member and get access to GitHub and Code:
/ allaboutai
🤖 Great AI Engineer Course:
scrimba.com/learn/aiengineer?...
🔥 Open GitHub Repos:
github.com/AllAboutAI-YT/easy...
📧 Join the newsletter:
www.allabtai.com/newsletter/
🌐 My website:
www.allabtai.com
Explaining the OpenAI GPT-4o API. My predictions and some tests of what I think we can expect from GPT-4o API and the multimodal model.
00:00 OpenAI GPT-4o API Intro
03:23 OpenAI GPT-4o Explained
06:47 OpenAI GPT-4o Exploration - Наука
It's going to be a game changer if OpenAI can actually deliver all the functions they demonstrated.
I just realized that new openai model killed your voice assistant projects, just as they did last time with GPTs
It’s a pattern I noticed and expect since gpt4 dropped.
or it can make it easier to build an with low latency, in the gpt ap you don't really have long term memory, or function calling, their is still a lot of room to make a great, personnalize local assistant imo + this open the way on omnicient llm, mabye in the futur local llm will have voice and vision natively
@@josephtilly258maybe but doubt this year, I can’t even run the 7b text modes and to add vision and audio understanding would need to be bigger
yes haha, but hey, thats technology and why we love it
This was very educational. Your instructions were clear and concise. ❤🎉
Hi Kris can you do it vision version also with camera ? Some things or help usecases
4:59 It's not horribly wrong, but I'd combine the and Voice IN in one graphic in the GPT-4o Voice API Now section and Voice OUT under the LLM RESPONSE. GPT-4o is going to be a game changer for education.
I wounder how they manage to handle interuptions during the voice output like in their demo for the api
I think you're right. Maybe minutes after OpenAI presentation was done, I posted on their developer forum if voice in / voice out will be available to developers soon. They said only to a small group of "trusted" partners. So yea, I'm not sure when we gonna get access to this. You gotta be in that special circle. 😅
What we need ASAP is an open source alternative to GPT-4o realtime speech-to-speech (as in demos). I'm pro open-source and I want full control of the application flow, preferably offline. Has anyone tried to use XTTS streaming capabilities succesfully, for example by extending AllAboutAI-examples?
Great content!
thnx :D appriciate it
Does the streaming audio function help with the latency?
You’re the best man.
thnx :D appriciate it
That "Performance Scores" table - nice! If that is all correct then that's pretty impressive.
Though I did a screenshot test myself and it mistook a 3 for an 8 so it might not be flawless.
yeah, i expect it to just improve over time until its perfect tho
How can I get this code?
❤❤❤
if you could give it IQ tests visually ! that would be great since this is difficult for it like how many triangles or what is the next humber in a series..
Hi, great video! Iam courios how much do those apis cost. On their website I found text pricing in tokens and it is pretty cheap and understandable. However the image or “vision” function seems to be so expensive. I calculated it and with full hd on low settings it is going to cost about 5$ for a minute on 15fps. That’s crazy, not even mentioning it their TTS, that costs 15$ per 1M tokens, which is pretty hilarious
I made a script just like yours and now with GPT-4o it kinda defeats the purpose...😅at least we can still use it with local models.
Hey Bro Is This Model Free To use, I Men Can We Explore This?
multimodal or multimodel ? does really anyone believe it’s a single model ?
My thoughts that we already have a good chance to be so so devs and to write perfect code 🎉😅
GPT5 may finally be able to tell how many r's are in the word "strawberry", but 4o will suffice with its ability to write a bigint from scratch in C in just a couple minutes of telling it to try again
Really ?
llma 3 8b even answer it ...
How many r's are in the word "strawberry"?
The word "strawberry" has 3 R's.
@@mirek190 ask it where they are. Also which Llama 3 8b model? Mine failed on first attempt, both meta and groq
@@mirek190 here's llama 3 70b failing: How many r's in strawberry
There are 3 R's in the word "strawberry".
Where are they?
I apologize for the mistake! There are actually 2 R's in the word "strawberry". They are consecutive, appearing as "rr" in the middle of the word.
@@mirek190 llama 3 70b groq also failed this
@@mirek190 and here's claude 3 opus:
The three r's in "strawberry" are located as follows:
1. st*r*awberry (after the first "t")
2. stra*w*berry (after the "w")
3. strawbe*r*ry (near the end, before the final "y")
You be the judge lmao
god the number of men who will fall in love with that voice hahaha
It will make the movie _Her_ seem like an underestimate.