GPT-4o Is Out, And It's Alive?!

Hey Creators! 👋

Did OpenAI just drop the mic on us again? When OpenAI's CTO, Mira Murati, announced the new multimodal GPT-4o yesterday, everyone’s jaws hit the floor. It features vision, real-time human-like speech, a response rate as fast as 232ms — and the kicker? It includes all the premium features of ChatGPT Plus (even the GPT store and plugins), and it’s free for everyone!

Hold onto your ChatGPT Plus subscriptions — in this post, we’ll cover:

• 100% Free & Much Better Than GPT-4? All the new & improved features of the newest multimodal GPT-4 “Omni”
• It’s Alive?! How GPT-4o’s vision & speech capabilities make it almost human (we’re living inside of the movie “Her”)
• GPT-4o Is Confusing? Every concern & question that arose from yesterday’s unexpected launch (why should we pay for ChatGPT Plus anymore?)
• How To Use It - The best new use cases & how to adapt to the latest features

The “o” Is For Omni

The "o" in GPT-4o stands for "Omni," highlighting its advanced multimodal capabilities. GPT-4o can seamlessly handle text, speech, and visual inputs, making it more versatile than previous models. It offers real-time, human-like responses with the lowest delay being 232ms, simultaneously processing text and audio inputs.

Additionally, GPT-4o is designed to work more efficiently in multiple languages, making it a powerful tool for global users. This upgrade means users get a faster, more integrated experience!

Here’s a little taste of just how fast it is at doing 3 moderately difficult tasks at once:

Although the GIF is sped up 150%, it took GPT-4o only about 13 or so seconds to generate a 4K image of a snowman dancing, a python calculator script, and a Pushkin-like poem in 2 languages — isn’t that insane?

Here’s the gist of the latest features & upgrades GPT-4o presents

1. Multimodal Capability - it can handle text, speech, and visual inputs, allowing for natural & seamless interactions.
2. Real-time Processing - responds as quickly as 232 milliseconds for speech & text output, which is much faster than its predecessor, GPT-4 Turbo.
3. Cost Efficiency - 50% cheaper in API costs compared to GPT-4 Turbo.
4. Global Usability - not only is it 100% free for everyone, but GPT-4o has enhanced language support for over 20 languages!
5. Paid Features Now Free - The GPT Plugin Store is now available to everyone with a ChatGPT account, no need for the Plus subscription anymore (we’ll complain about this later in the post)
6. Mac Desktop App - GPT-4o launched with a Mac Desktop app, with which you can capture meetings, share your screen for an integrated experience & more.

Let’s break down some of the improvements in a little more depth.

Do you have a minute to participate in our Subscription Giveaway and make our newsletter even better? Take a quick survey and tell us about your experience & participate in the free Giveaway of a Premium Subscription for 6 months.

Join The Giveaway

It’s Way Too Real

So it can see, listen to you, and talk back to you like a human, all in real time?

Vision, Speech & Human Capabilities

We’re actually so close to living in the movie “Her”. It sounds way too human to be true, you can even make it blush?!

These vision capabilities open up so many doors for use, such as asking it whether you look presentable for an interview, having a quick chat or getting thoughts on a new suit you’re buying!

Math Problems & Education

GPT-4o doesn’t just solve math problems now; it does so with a touch of personality & care. The man on the right of the GIF is the founder of Khan Academy, a popular online education platform.

When he interrupted the AI at the start, it (she) responded with “Tsk, of course, I'd be happy to help.”, slightly annoyed but ever-ready to assist. It’s like having a sassy but caring tutor!

The Impact GPT-4o Will Have On Education

Now, anyone, regardless of location, financial situation, or previous education, can have an extremely intelligent tutor. GPT-4o offers infinite patience and personalized help, all for free. Students worldwide can now access high-quality education, leveling the playing field and giving absolutely no one the excuse not to excel in their studies.

RIP to tutors and teachers — GPT-4o is here to support students like never before, and it’s here to stay.

Share this with your friends and colleagues, especially if they are AI-savvy!

Share

This Changes Everything

Those were just two previews of the capabilities, amongst the more niche yet life-changing capabilities are:

1. Helping the blind navigate without guide dogs (GPT-4o can be your eyes & ears!);
2. Translate accurately in real-time, effectively breaking the worldwide language barrier;
3. Code alongside you (screen share your IDE and let it do the work “with you”);
4. It can even crack dad jokes & sing you happy birthday — all in a pretty voice, just like a human would;

GPT-4o can even compliment your dog & it’s ball-playing skills, I still can’t wrap my head around how this just released yesterday.

What are your thoughts on GPT-4o? What will you use it for when you get your hands on it first? Leave a comment and let us know!

Leave a comment

Why We’re Confused

The unexpected release of GPT-4o left us with many questions.

With GPT-4o offering so much for free, we’re all questioning the value of a ChatGPT Plus subscription. We get the GPT Plugin store, built-in web browsing, and 40 messages per hour — all in the free GPT-4o version, while the premium Plus plan just grants us a 5x larger messaging limit?

It’s safe to say there isn’t much of a need for paying for ChatGPT Plus anymore until OpenAI gives us subscribers more goodies.

Who even sends ChatGPT 200 messages every 3 hours? Even the previous limit of 40 msg/3hr limit of the Plus plan was enough to get most of my content creation & data analysis done!

Among these questions, it's clear that GPT-4o, impressive as it is, is still in its early stages. It can sometimes "become a robot" mid-conversation or have voice breaks. It might also make mistakes with code or provide inaccurate information.

Additionally, the Windows app won’t be available until later this year. Meanwhile, Windows users will have to rely on Copilot and Bing, while Apple users enjoy early access to real-time voice interactions and screen-sharing features.

Now for some applicable information, we’ll showcase some of the best new use cases for GPT-4o and a very special prompt formula to get high-quality outputs from GPT-4o consistently.

The Best Use Cases For GPT-4o & How To Use It Properly

Let’s start off with the newest uses for GPT-4o that weren’t possible with the previous model, and our special formula to squeeze the most out of GPT-4o and become a “prompt engineer” yourself!

Get more from Creators' AI in the Substack app

Available for iOS and Android

Get the app

Your Own Personal Meeting Assistant

With the Mac Desktop GPT-4o app, you can “invite” GPT-4o to your meetings. This has a few uses, such as remembering important information to create a summary at the end, chatting with the meeting members, and giving insights in real time.

We went into extreme detail of the prompt formula in one of our posts here:

Here’s the full prompt formula:

Context + Details + Role + Intent + Response Format = Golden Prompt

And this is how you should prompt with GPT-4o during a meeting:

1. Set The Context - introduce who you & your meeting members are, where you work, and what your & everyone else’s roles are.

2. Include All The Details - give the specifics of the meeting, and state exactly why this meeting is taking place. Include all metrics & statistics you have so that GPT-4o can accurately respond.

3. Role of GPT-4o - assign GPT-4o a role. If the meeting’s purpose is to increase sales & customer LTV, tell it to act as an expert-level salesman & operations specialist.

4. Clarify The Intent - tell GPT-4o what you want it to do. The intent has to be clear, in the sales example, ask it to give actionable steps to take to increase sales & the lifetime value of customers.

5. Response Format - this is more useful in text output, but in the voice use case, you can ask it to respond in a step-by-step format, otherwise it will just talk for minutes in long paragraphs.

Remember, this is just an example. You can ask it to record the meeting & remember important parts, give insights only when asked to, or help cheer up the members — make sure to alter the above formula appropriately.

Creators’ AI Subscription is the best gift for Creators diving into AI.

Give a gift subscription

Real-time Coding Assistance

Using the Mac Desktop App, you can share your screen and allow GPT-4o to coach you with coding. It’s like having your own personal programmer, but you learn & code at the same time. Here’s an example of our special formula at play within this scenario:

1. I am currently writing a game in C++. I have been studying C++ for just under 3 months, and my knowledge is limited.

2. Currently, I’m trying to write the code for an NPC to respond when the player interacts with it. I want the NPC to respond with 3 dialogue options, positive, negative & neutral, depending on the player’s dialogue choice.

3. I want you to act as an expert-level C++ game designer & programmer and assist me in coding.

4. Help me code both the NPC and the player’s dialogue and the interaction between the two.

5. I want you to ask me questions and help me learn along the way, don’t just tell me what to do. Respond in a caring & patient format.

Your Own Personal Language Tutor

You can use GPT-4o in real-time to teach you a language. Let’s engineer the prompt as if you want to learn Spanish:

1. I am currently at level B2 in Spanish, but I want to reach C1 as quickly as possible.

2. I’m trying to read this book, and it’s a little above my level. There are some words that I can’t understand.

3. Act as an expert-level Spanish tutor and help me understand everything on this page.

4. I will read a sentence, and afterward, you will tell me where my intonation and pronunciation were bad, and what’s correct.

5. Then, you give me an accurate translation back to English and ask if I have any questions & address them.

With the above prompt formulation, you can read a book alongside GPT-4o and have it tutor you in sync — for free.

Why pay for a language tutor when GPT-4o has unlimited patience and can teach you a language for hours on end?

The above is just the tip of the iceberg of possibilities with this model! And as long as you use our special prompt formula, you can leverage the power of GPT-4o to its fullest:

Context + Details + Role + Intent + Response Format = Golden Prompt

That’s that, creators! In this post, we covered:

• The New Capabilities of GPT-4o
• Human-like Speech, Vision & Quick Responses
• The Confusion GPT-4o Created
• The Best Use Cases + Our Special Prompt Formula In Action

What do you think of GPT-4o? Have you tried it yet? Give us your thoughts in the comments!

Leave a comment