OpenAI's Mini Model Makes Big Waves

Wow, OpenAI is moving at lightning speed!

It's hard to keep up, right?

Just when I thought there was just one thing cooking up, they surprised me again. Remember that "Strawberry" project we talked about?

It's still being tested, but it's already showing signs of some pretty impressive thinking skills.

But you have to hold your breath because they have come up with the most “capable and cost-efficient model” available today.

They've got even bigger plans in the works.

As an AI enthusiast, one piece of news isn't enough to satisfy my curiosity.

There is a lot more on what OpenAI is cooking up next!

So let’s dive for amazing insights on OpenAI.

What’s Inside Today’s Newsletter

🪰 Buzz Around AI

⚙️ AI Tools Highlight

🫧 AI Creations

🪄 Prompt Of The Week

🔹 AI Learning

♨️ What’s Hot In AI

🌐 AI News

Read time: 5 mins

🪰 Buzz Around AI

OpenAI launched a new model, GPT4o-mini

You might have a cousin who behaves in the same way as their parents do, but they're just a smaller version of their parents.

Similarly, today's news that I'm going to cover is about GPT-4o mini.

It's like the smaller, faster cousin of OpenAI's big AI model. This little guy is smart and cheap to run.

It's so good that it'll soon power the free version of ChatGPT.

But you might be wondering why smaller version when we have the original one:

So let me tell you that all the big tech companies are making these smaller AI models now.

But why? It's all about money.

The big models cost a ton to run. But these mini versions? They're way cheaper.

You can relate it to a smartphone. Remember when only rich people had smartphones? Now everyone has one.

That's what's happening with AI. These smaller models make AI cheaper, so more people can use it.

🚀 What's really exciting about this is:

🏆 OpenAI says this new model is the best and cheapest small AI out there, and they've got proof to back it up.

🆓 Soon, when you use the free ChatGPT, you'll be chatting with this new AI instead of the older one.

🔒 They added a new safety trick that makes it harder for users to make the AI do things it shouldn't.

💰 This new AI costs way less to run than the old one - it's 60% cheaper! And get this - it's actually smarter too! 🧠

🔹AI Learning

🎬Create Mind-Blowing Lip-Synced Videos with RunwayML Gen-3

🔑 Log into RunwayML's site to grab some credits

🚀 Hit that "Try Lip Sync" button (make sure you've got Gen-3 access!)

🎭 Time to play director:

• Write your script ✍️

• Add a voice 🗣️

• Pick a face (if you need one) 😎

• Click on that "Generate" button!

⏳ Grab a coffee while the AI works its magic

🎉It’s Ready!!

Download your perfectly lip-synced masterpiece

♨️ What’s Hot In AI

OpenAI vs. Nvidia: The AI Chip Battle Heats Up 🔥

Imagine a race car driver always borrowing engines. They're slow to arrive and cost a fortune.

One day, he thinks: "I'll build my own." It takes time, but soon he got a custom powerhouse.

OpenAI wants to do the same with chips.

Like other AI giants, they use Nvidia's semiconductors. But their supply is tight. OpenAI worries they won't have enough for GPT-5 and beyond.

They're talking to chip makers like Broadcom to build its own NPUs.

Alphabet did this before. They made TPUs for Gemini.

OpenAI has hired Alphabet's TPU experts.

It's not OpenAI's first try. Sam Altman once sought $7 trillion to reshape the chip industry.

The challenge is steep as Nvidia is about five to 10 years ahead of the competition.

If OpenAI partners with a chip maker, results will take years. But they may have no choice.

An insider said, "AI's limit is capacity: chips, energy, compute. OpenAI won't let others build that. They're on the front line."

OpenAI is taking control. Like the race car driver with her custom engine, they're building their own future.

🫧 AI Image Of The Day

Prompt:- a muscle car, galaxy brainiac GTI edition, magazine advertisement quality 35mm Kodachrome photograph of a galaxy muscle car with glass headlights, at night with wet pavement and photorealistic blurred city night background. --ar 16:11

⚙️ AI Tools Highlight

🔍 Blobr: Your AI sidekick that sniffs out hidden customer trends across all your apps.

It's like having a super-smart detective on your team!

🧠 IBM's Watsonx: The Swiss Army knife of AI platforms. Train, tweak, and launch AI models using your own secret sauce data.

📣 Content Codex: Your personal marketing guru powered by AI. It whips up strategies tailored just for your brand and serves you insights on a silver platter.

🚀 Heyday: Think of it as your AI-powered memory bank on steroids.

It turns your docs, notes, and chats into a goldmine of quotes, shareable nuggets, and a database you can quiz anytime. It's like having a super-organized assistant with a photographic memory!

🪄 Prompt Of The Week

Prompts to learn ANYTHING 10x faster:

You can copy and paste this prompt into any Chatbot

Use the 80/20 principle to learn faster:

"I want to learn about [insert topic]. Identify and share the most important 20% of learnings from this topic that will help me understand 80% of it."

Source @amkhanniazi

🌐 AI News

• Apple has just launched an exciting new AI model called DCLM-7B.

• Google Launches Project Oscar To Streamline Software Development.

• Luma Labs released a new Loop feature for Dream Machine videos.

• Elon Musk has begun training xAI with 100,000 NVIDIA H100 GPUs, which is the world's most powerful AI training cluster.